Self-contained audio and annotation gallery · built 2026-10-03 10:34 UTC

Hear what the Whisper multitask models predict

Thirty clips: twelve generated predictions from three listening studies, twelve seeded Emolia/Emilia samples in six languages, and six curated P3 test clips with inserted vocal bursts. The same audio is annotated by completed Whisper Tiny and Whisper Base checkpoints. Each player contains Base64 audio, so this is one self-contained file.

Model weights, inference and training code · Scaling Ladder dataset

Training-reference z-scores. z = (raw model prediction − training teacher median) / training teacher standard deviation. The first 100 targets use all 100,858 S9 Ladder t0+t1 clips. The additional 92 use a seeded 100,000-example P3 training subset. z shows how unusual a score is for that dimension; it is not a probability, preference or human rating. Interest and Concentration no longer dominate just because of high raw baselines. These display statistics do not change checkpoint inference.

What is shown

All 192 scalar outputs per model, grouped into emotions, VoiceNet, genuineness, blend, Empathic Insight Plus, AudioBox, DNSMOS, burst count and VoiceCLAP attributes. R_quality was trained from Ladder but has no P3 test labels. Character rate (CPS) is computed from source metadata when available; it is not a learned head. Raw VoiceCLAP 768D and BUD-E 3072D embeddings were stored as training sidecars but are not outputs of these checkpoints.

Speaker vectors

Each checkpoint predicts Orange timbre (128D) and identity (250D). A cosine of 1 means identical direction to the Orange teacher vector; values closer to 1 indicate stronger reconstruction. The full vectors are expandable below each clip.

Burst times

The frame head predicts a value every 20 ms. Bars and the start/end list use a fixed 0.5 threshold with no smoothing. The separate count head is a regression estimate and may differ from the number of contiguous bars.

Dataset scope

Generated examples are from the three supplied HTML studies. Natural examples are two seeded Emolia clips per language from six shards; they may occur in training. The six P3 clips are from the source-disjoint test split but deliberately selected for distinctive teacher emotion scores and inserted bursts. This curated gallery is for interpretation, not a held-out performance estimate. Consult the separate test report in the model repository.

Showing all 30 clips

g01EnglishDPO acting study

HVSAC-001 · en · targets: Doubt, Sadness, Triumph

Generated TTS prediction · Rank 16 × 0.50 · C01 · 19.7 s

Grandma... I am beside your hospital bed, holding the sealed letter you gave me. I thought it would explain the secret... instead it feels like you trusted me with your whole past. I wish I could ask what to do, but I will carry this truth forward.

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Thankfulness Gratitude
z +1.45 / +2.49raw 1.05 / 1.78
Relief
z +0.97 / +1.76raw 0.70 / 1.23
Sadness
z +1.73 / +1.55raw 1.31 / 1.17
Shame
z +1.65 / +1.24raw 0.85 / 0.65
Disappointment
z +1.46 / +1.29raw 1.02 / 0.90
Distress
z +1.42 / +1.26raw 1.16 / 1.03
Contemplation
z +1.00 / +1.32raw 0.68 / 0.87

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Authoritative style (S_AUTH)
z +1.69 / +2.13raw 2.85 / 3.31
Fullness (FULL)
z +2.05 / +1.98raw 3.60 / 3.54
Content appropriateness (EXPL)
z +1.88 / +1.93raw 0.69 / 0.71
Narrator style (S_NARR)
z +1.91 / +1.74raw 4.07 / 3.83
Esthetics (ESTH)
z +1.85 / +1.61raw 3.85 / 3.66
Chest resonance (R_CHST)
z +1.67 / +1.67raw 3.55 / 3.55
Mixed resonance (R_MIXD)
z +1.53 / +1.34raw 2.72 / 2.56

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z -1.06 / -1.10raw 0.62 / 0.55
Vocal-burst blend (0–10)z -1.00 / -1.31raw 3.95 / 3.08
Content enjoyment (CE)z +0.35 / +0.44raw 5.16 / 5.24
Production quality (PQ)z +0.34 / +0.12raw 7.34 / 7.15
Speech signal quality, calibratedz +0.11 / +0.23raw 3.54 / 3.60
Background quality, calibratedz +0.54 / +0.68raw 3.89 / 4.02
Overall quality, calibratedz +0.55 / +0.79raw 3.18 / 3.32
P.808 MOS estimatez +0.86 / +1.05raw 3.85 / 3.94
score content enjoymentz +0.72 / +1.10raw 4.72 / 4.84
score overall qualityz +0.08 / +0.63raw 2.69 / 2.97
Background Noisez -2.04 / -2.88raw 0.27 / 0.05
Tempo (TEMP)z +0.12 / +0.17raw 2.26 / 2.32
Arousal (AROU)z -0.24 / +0.04raw 2.08 / 2.40
Valence (VALN)z -0.30 / -0.49raw 1.99 / 1.78
Recording quality (RCQL)z +0.75 / +0.68raw 3.42 / 3.36
Background noise (BKGN)z +1.07 / +0.82raw 3.51 / 3.33
talking speedz +1.36 / +1.32raw 14.14 / 13.97
Characters per second (CPS): 12.59 raw · z +0.95

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.7610.806
Identity · 250D0.6980.680

Vocal-burst regions

Tiny
Base
0 s19.7 s

Verified burst intervals are unavailable for this clip.

Tiny: 0 regions, count-head estimate 0.00; Base: 0 regions, count-head estimate 0.01.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.698+1.210.575+1.000.0070.570
1Amusementemo_Amusement0.174+0.20-0.095-0.11-0.0010.885
2Angeremo_Anger0.495+0.800.588+0.950.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.485+0.840.314+0.540.0050.572
4Aweemo_Awe0.253+0.590.039+0.090.0000.427
5Bitternessemo_Bitterness0.434+1.010.192+0.430.0130.415
6Concentrationemo_Concentration0.810+0.241.216+0.780.6340.741
7Confusionemo_Confusion0.450+0.830.447+0.830.0010.540
8Contemplationemo_Contemplation0.683+1.000.869+1.320.0820.598
9Contemptemo_Contempt0.167+0.36-0.019-0.04-0.0000.464
10Contentmentemo_Contentment0.426+0.830.378+0.740.0000.514
11Disappointmentemo_Disappointment1.024+1.460.903+1.290.0000.700
12Disgustemo_Disgust0.346+0.690.026+0.040.0070.489
13Distressemo_Distress1.165+1.421.034+1.260.0000.821
14Doubtemo_Doubt0.252+0.530.394+0.830.0060.466
15Elationemo_Elation0.131+0.160.011-0.000.0120.728
16Embarrassmentemo_Embarrassment0.327+0.660.374+0.770.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.683+0.740.592+0.640.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.174+0.270.044+0.030.0280.539
19Fearemo_Fear0.625+1.000.514+0.820.0150.610
20Helplessnessemo_Helplessness0.885+1.280.824+1.19-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.429+0.410.289+0.230.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.437+0.410.477+0.470.1490.705
23Infatuationemo_Infatuation0.445+0.860.288+0.57-0.0150.533
24Interestemo_Interest1.790+0.221.658+0.081.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness-0.022+0.01-0.040-0.03-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.255+0.440.088+0.060.0590.442
27Longingemo_Longing0.890+1.270.623+0.850.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.529+0.890.335+0.520.0550.534
29Painemo_Pain0.673+1.160.402+0.70-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.301+0.460.090+0.120.0170.614
31Prideemo_Pride0.211+0.410.442+0.870.0020.509
32Reliefemo_Relief0.696+0.971.231+1.760.0330.681
33Sadnessemo_Sadness1.313+1.731.172+1.550.0000.759
34Sexual Lustemo_Sexual_Lust0.353+0.650.049+0.040.0300.494
35Shameemo_Shame0.854+1.650.647+1.240.0130.510
36Sournessemo_Sourness0.399+0.920.050+0.13-0.0050.439
37Teasingemo_Teasing0.333+0.610.026+0.050.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude1.053+1.451.777+2.490.0410.697
39Triumphemo_Triumph0.110+0.220.230+0.46-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg3.164+0.143.375+0.322.9841.238
41Arousal (AROU)vn_AROU_reg2.085-0.242.403+0.042.3591.147
42Arousal shift (ARSH)vn_ARSH_reg2.369-0.242.739+0.182.5800.867
43Vocal attack (ATCK)vn_ATCK_reg2.162+0.002.219+0.042.1621.295
44Background noise (BKGN)vn_BKGN_reg3.515+1.073.326+0.822.7030.761
45Brightness (BRGT)vn_BRGT_reg2.372+0.472.331+0.421.9910.813
46Speech chunking (CHNK)vn_CHNK_reg2.568+0.582.525+0.551.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg2.529+0.542.735+0.762.0180.945
48Cognitive load (COGL)vn_COGL_reg1.437-0.621.260-0.792.0941.061
49Dynamic arc (DARC)vn_DARC_reg1.139-0.351.142-0.341.6151.380
50Disfluency (DFLU)vn_DFLU_reg0.914-1.330.951-1.302.4881.184
51Emphasis (EMPH)vn_EMPH_reg3.289+0.453.129+0.342.6071.524
52Esthetics (ESTH)vn_ESTH_reg3.852+1.853.663+1.612.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.692+1.880.710+1.930.0860.322
54Focus (FOCS)vn_FOCS_reg3.243+0.823.117+0.702.3441.102
55Fullness (FULL)vn_FULL_reg3.601+2.053.539+1.981.7780.889
56Perceived gender (GEND)vn_GEND_reg4.883+0.525.076+0.644.0551.588
57Harmonicity (HARM)vn_HARM_reg2.219-0.312.253-0.272.4800.836
58Metallic character (METL)vn_METL_reg1.235-0.401.343-0.291.6210.958
59Pitch range (RANG)vn_RANG_reg2.178-0.822.174-0.822.9140.897
60Recording quality (RCQL)vn_RCQL_reg3.416+0.753.360+0.682.7540.888
61Vocal register (REGS)vn_REGS_reg0.790-0.320.750-0.351.1981.268
62Respiration (RESP)vn_RESP_reg2.207-0.451.856-0.782.6801.056
63Roughness (ROUG)vn_ROUG_reg2.160+0.142.071+0.042.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg3.547+1.673.547+1.672.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg1.630-0.631.703-0.542.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg1.139-0.761.477-0.431.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg2.723+1.532.560+1.341.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg2.503-0.272.457-0.322.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg1.967-0.292.064-0.162.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg1.991-0.102.003-0.082.0700.803
71Smoothness (SMTH)vn_SMTH_reg3.044+0.822.966+0.752.1431.093
72Stance (STNC)vn_STNC_reg3.025+0.353.210+0.512.6331.137
73Speech structure (STRU)vn_STRU_reg2.869+0.763.108+1.012.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg1.796+0.301.199-0.121.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg2.851+1.693.305+2.131.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg1.190-0.261.011-0.391.5601.401
77Casual style (S_CASU)vn_S_CASU_reg2.569-0.962.415-1.083.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg2.516-0.062.821+0.172.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg2.820+0.092.798+0.072.7211.148
80Formal style (S_FORM)vn_S_FORM_reg2.569+1.352.624+1.401.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg3.824+0.783.571+0.582.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg4.065+1.913.830+1.741.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.881+0.831.043+1.120.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg1.024-1.010.857-1.142.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg0.203-0.780.819-0.311.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg4.236+1.294.323+1.352.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg1.147+0.410.789+0.120.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg3.878+1.413.216+1.001.5971.618
89Tempo (TEMP)vn_TEMP_reg2.262+0.122.319+0.172.1171.190
90Tension (TENS)vn_TENS_reg2.105+0.642.287+0.861.5590.852
91Valence (VALN)vn_VALN_reg1.993-0.301.782-0.492.3281.118
92Valence shift (VALS)vn_VALS_reg2.374-0.132.309-0.192.5181.079
93Velocity flux (VFLX)vn_VFLX_reg1.057-0.991.052-1.002.2561.208
94Volatility (VOLT)vn_VOLT_reg2.098-0.551.877-0.802.5860.886
95Vulnerability (VULN)vn_VULN_reg2.051-0.611.548-1.122.6480.984
96Warmth (WARM)vn_WARM_reg3.892+1.443.602+1.132.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_60.616-1.060.553-1.102.3321.621
98Vocal-burst blend (0–10)blend_0_103.947-1.003.077-1.316.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.510-1.730.412-2.021.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.218-0.113.270+0.073.2500.293
101Arousaleiv_extra:Arousal1.560+0.621.240-0.111.2890.435
102Authenticityeiv_extra:Authenticity2.963+1.942.923+1.682.6670.153
103Background Noiseeiv_extra:Background_Noise0.268-2.040.047-2.880.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.160+0.401.217+0.591.0400.303
105Gendereiv_extra:Gender-0.744-0.12-1.145-0.53-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.991+0.652.091+1.051.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive2.059+0.422.004+0.311.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.577+1.312.648+1.522.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous0.793-0.400.728-0.590.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.444+0.510.410+0.410.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.793+1.290.693+0.980.3760.324
112Valenceeiv_extra:Valence-0.291-0.78-0.215-0.580.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.708-0.011.765+0.121.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold-0.043-0.31-0.021-0.240.0510.300
115score background qualityeiv_extra:score_background_quality3.410+0.163.616+0.453.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.724+0.724.840+1.104.5070.303
117score overall qualityeiv_extra:score_overall_quality2.692+0.082.971+0.632.6520.507
118score speech qualityeiv_extra:score_speech_quality1.801-0.051.844+0.251.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.159+0.355.242+0.444.7991.018
120Content usefulness (CU)audiobox:CU6.464+0.816.401+0.755.6491.001
121Production complexity (PC)audiobox:PC1.883-0.281.549-0.482.3311.617
122Production quality (PQ)audiobox:PQ7.339+0.347.148+0.127.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw4.010+0.164.116+0.323.9000.669
124Background quality, rawdnsmos:BAK_raw4.109+0.804.232+0.913.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.630+0.673.840+0.933.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.542+0.113.605+0.233.4880.501
127Background quality, calibrateddnsmos:BAK3.887+0.544.017+0.683.3890.928
128Overall quality, calibrateddnsmos:OVRL3.176+0.553.319+0.792.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.847+0.863.942+1.053.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p-0.008-0.010.013+0.020.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection1.001+0.101.175+0.410.9470.554
132Agevoiceclap_attribute:Age3.061-0.313.309+0.153.2290.547
133Amusementvoiceclap_attribute:Amusement0.299-0.160.360-0.060.3920.572
134Angervoiceclap_attribute:Anger1.228+0.471.370+0.750.9920.506
135Arousalvoiceclap_attribute:Arousal1.536+0.191.731+0.501.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.735+0.470.870+0.750.5090.479
137Authenticityvoiceclap_attribute:Authenticity3.105+2.003.148+2.262.7700.167
138Awevoiceclap_attribute:Awe0.751+0.820.839+1.160.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.177-1.480.234-1.260.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.653+0.250.737+0.610.5920.236
141Concentrationvoiceclap_attribute:Concentration1.204-0.021.116-0.181.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant0.939-0.281.034-0.081.0690.468
143Confusionvoiceclap_attribute:Confusion0.447-0.590.481-0.490.6430.334
144Contemplationvoiceclap_attribute:Contemplation1.049+0.660.877+0.240.7800.406
145Contemptvoiceclap_attribute:Contempt0.799+0.270.870+0.480.7090.339
146Contentmentvoiceclap_attribute:Contentment0.709+0.390.571-0.060.5880.307
147Disappointmentvoiceclap_attribute:Disappointment1.060+1.031.131+1.270.7600.291
148Disgustvoiceclap_attribute:Disgust0.539-0.410.679+0.260.6250.207
149Distressvoiceclap_attribute:Distress1.207+0.801.391+1.330.9270.350
150Doubtvoiceclap_attribute:Doubt0.521-0.070.602+0.360.5350.188
151Elationvoiceclap_attribute:Elation0.655+0.290.694+0.390.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.275-0.100.340+0.110.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.839+0.660.885+0.830.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.490+0.110.420-0.160.4620.261
155Fearvoiceclap_attribute:Fear0.895+1.310.950+1.570.6180.212
156Gendervoiceclap_attribute:Gender-0.767-0.61-0.940-0.78-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness1.017+1.341.040+1.420.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched2.120+0.752.132+0.781.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.917+0.541.005+0.760.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability0.878-0.321.165+0.211.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.554-0.160.768+0.480.6080.332
162Interestvoiceclap_attribute:Interest1.476-0.281.402-0.451.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness-0.070-1.240.093-0.880.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.649+0.520.920+1.620.5230.245
165Longingvoiceclap_attribute:Longing0.784+1.120.808+1.230.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.918+0.250.944+0.320.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive1.983+0.351.906+0.221.7750.594
168Painvoiceclap_attribute:Pain0.528+0.670.652+1.240.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.612+0.360.680+0.590.5030.301
170Pridevoiceclap_attribute:Pride0.988+0.591.346+1.620.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.876+1.722.776+1.412.3130.328
172Reliefvoiceclap_attribute:Relief1.313+0.701.515+1.181.0250.415
173Sadnessvoiceclap_attribute:Sadness1.222+1.501.266+1.640.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.572-0.410.483-0.560.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.449-0.440.420-0.510.6350.419
176Shamevoiceclap_attribute:Shame0.960+1.531.103+2.040.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.303+0.400.054-0.290.1600.360
178Sournessvoiceclap_attribute:Sourness0.715+0.030.798+0.320.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.922+0.830.950+0.910.6230.359
180Teasingvoiceclap_attribute:Teasing0.597+0.150.673+0.350.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.813+0.811.844+0.871.3840.529
182Triumphvoiceclap_attribute:Triumph0.988+0.761.076+1.000.7160.359
183Valencevoiceclap_attribute:Valence-0.113-0.17-0.415-0.64-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached1.920+0.512.009+0.671.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold0.112+0.45-0.030+0.09-0.0660.400
186durationvoiceclap_attribute:duration11.444+0.0812.331+0.2210.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.550+0.613.700+0.913.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.862+1.404.915+1.764.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.870+0.572.956+0.782.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.925+0.501.893+0.031.8910.069
191talking speedvoiceclap_attribute:talking_speed14.143+1.3613.967+1.329.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.06910.0211-0.0450
1-0.0576-0.0845-0.0794
2-0.0744-0.0543-0.0784
30.07710.06160.1256
4-0.0230-0.0226-0.0457
5-0.0141-0.0158-0.0266
60.08850.09360.1438
7-0.0702-0.0737-0.0731
80.05490.05250.0999
90.03720.07030.1221
100.00950.03460.0981
110.00480.06880.1649
12-0.0607-0.04620.0525
13-0.0052-0.01300.0238
14-0.0771-0.04500.0221
15-0.0557-0.0942-0.0716
16-0.0801-0.1183-0.0989
17-0.0018-0.0208-0.0589
180.06410.0367-0.0709
190.0219-0.0531-0.0641
200.02020.0046-0.0800
210.08430.04350.1161
22-0.1236-0.0564-0.0751
23-0.0089-0.01470.0688
24-0.0630-0.0881-0.0645
25-0.0855-0.0809-0.0702
26-0.04540.0026-0.0188
27-0.0774-0.0845-0.0745
280.08050.06340.0980
29-0.1351-0.0869-0.0856
300.18010.0634-0.0078
310.39070.39120.2549
32-0.0885-0.0924-0.0770
330.03590.07270.0933
34-0.0790-0.0459-0.0717
35-0.01090.01290.0792
36-0.0694-0.0851-0.0859
370.12510.0241-0.0134
380.0166-0.01220.0524
39-0.1152-0.1044-0.0806
40-0.0372-0.01840.0240
41-0.0557-0.0721-0.1149
420.0185-0.0291-0.0729
43-0.0568-0.0767-0.0822
44-0.0021-0.0513-0.0767
45-0.0437-0.0447-0.0516
46-0.0572-0.0930-0.0558
47-0.0364-0.0543-0.0808
480.0135-0.0120-0.0702
49-0.1045-0.0845-0.0454
500.09420.12010.1969
51-0.0138-0.07270.0269
520.03260.0308-0.0307
53-0.0992-0.0930-0.0815
540.05300.03740.0224
550.07780.15270.0919
560.09730.08150.1689
570.0173-0.00390.0100
58-0.1061-0.0845-0.0850
590.23650.21730.2803
60-0.1206-0.1322-0.0771
610.0186-0.07520.0222
62-0.1122-0.1123-0.0747
63-0.02410.0267-0.0759
640.07820.14250.0364
65-0.0546-0.0492-0.0652
660.0319-0.0037-0.0136
67-0.1007-0.0408-0.0183
680.03870.06850.0794
69-0.0671-0.0694-0.0732
70-0.1030-0.0779-0.0828
71-0.01430.01770.0522
72-0.0641-0.0839-0.0459
73-0.1251-0.0857-0.0977
74-0.0557-0.06100.0223
75-0.01310.0210-0.0970
76-0.0912-0.0779-0.0775
770.05380.13280.0425
78-0.0086-0.0417-0.0461
79-0.0740-0.0900-0.0693
800.13730.19560.1107
810.06100.04040.0525
820.02230.09180.0348
83-0.1259-0.0863-0.0818
840.04880.1220-0.0196
85-0.02200.06250.0359
86-0.0286-0.0131-0.0470
87-0.0012-0.0026-0.0033
88-0.0061-0.04740.0259
890.13960.09360.1165
90-0.1160-0.1582-0.0866
910.19000.12440.1757
92-0.1496-0.0924-0.0201
93-0.0717-0.0589-0.0781
940.01330.0264-0.0480
95-0.0935-0.0906-0.0858
960.11060.18720.0992
97-0.0205-0.0338-0.0854
980.04770.06640.1279
99-0.0618-0.1075-0.0793
100-0.0303-0.0318-0.0848
1010.09270.11950.1438
1020.07400.10080.1376
103-0.0599-0.0115-0.0844
104-0.0004-0.0029-0.0781
105-0.0942-0.0954-0.0770
106-0.02560.0305-0.0961
1070.02040.07670.0771
1080.04830.0501-0.0866
109-0.0801-0.0857-0.0784
1100.13350.13820.2197
111-0.0599-0.0803-0.0927
112-0.1106-0.1189-0.0788
1130.04880.00130.0237
114-0.1267-0.0978-0.0861
115-0.1068-0.0930-0.0757
116-0.0683-0.0869-0.0685
1170.27320.21370.1322
118-0.0866-0.0734-0.0778
1190.04220.0134-0.0074
120-0.0862-0.1069-0.0470
121-0.1007-0.1020-0.0810
122-0.03870.05010.0493
1230.0209-0.0094-0.0509
1240.0265-0.0152-0.0883
125-0.0656-0.0567-0.0797
126-0.0729-0.1044-0.0688
127-0.1213-0.0954-0.0848
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
00.00740.0215-0.0009
1-0.1012-0.1056-0.0408
2-0.0888-0.0392-0.0518
3-0.1312-0.1193-0.0800
4-0.0176-0.0100-0.0384
5-0.0502-0.0397-0.0341
60.02150.01330.0061
70.07760.0226-0.0117
8-0.02020.0402-0.0272
90.18590.15660.1617
100.10180.08730.0523
11-0.0489-0.0066-0.0083
12-0.01390.0141-0.0158
13-0.0410-0.0410-0.0372
14-0.0472-0.0251-0.0509
150.03950.12190.0723
160.06150.0773-0.0098
170.05400.03650.0939
18-0.0601-0.0449-0.0429
19-0.01990.03070.0381
20-0.0366-0.0183-0.0200
21-0.0379-0.0631-0.0391
22-0.0143-0.0507-0.0168
23-0.0731-0.0036-0.0408
24-0.01660.0044-0.0360
250.0008-0.0025-0.0455
26-0.0328-0.0673-0.0465
27-0.0070-0.0071-0.0343
28-0.0871-0.0694-0.0457
290.0077-0.0101-0.0406
30-0.0253-0.0363-0.0367
31-0.0314-0.0620-0.0530
32-0.0533-0.0307-0.0434
330.11960.03630.0891
34-0.03200.0191-0.0504
350.01160.07990.1178
36-0.1668-0.0678-0.0446
37-0.0032-0.0357-0.0292
38-0.1422-0.1598-0.0585
39-0.0584-0.0164-0.0231
40-0.0275-0.05970.0332
41-0.0091-0.02610.0111
420.03280.06390.0331
430.01500.0507-0.0016
440.01520.03360.0193
45-0.0639-0.0272-0.0484
46-0.0151-0.0256-0.0149
470.08370.05470.1667
48-0.0141-0.0332-0.0446
490.0036-0.02890.0606
500.02480.0247-0.0345
510.09160.04570.0936
52-0.0149-0.0363-0.0416
53-0.00900.0029-0.0408
540.07690.0139-0.0144
55-0.0694-0.0339-0.0675
56-0.0227-0.0549-0.0247
57-0.0119-0.0213-0.0492
580.06150.0217-0.0067
590.05710.07040.0519
60-0.0649-0.0699-0.0442
610.02460.03070.0318
62-0.00880.03340.1034
63-0.0105-0.03990.0972
64-0.00740.0118-0.0315
650.21600.15660.2203
66-0.0895-0.0562-0.0507
670.01750.02510.1021
68-0.0002-0.01630.0495
69-0.0132-0.0188-0.0324
70-0.0502-0.05020.0538
71-0.0302-0.00380.0491
72-0.0412-0.0736-0.0467
730.0179-0.05390.0465
74-0.0502-0.0725-0.0455
750.11280.13090.1020
76-0.0183-0.0293-0.0402
770.04030.06520.0429
78-0.0120-0.02140.0262
790.07960.11720.0973
80-0.02560.04570.0467
810.05570.06780.1071
820.03670.01610.0224
83-0.0537-0.0544-0.0374
84-0.0584-0.0141-0.0635
85-0.0496-0.0618-0.0450
86-0.0043-0.0017-0.0402
87-0.0759-0.0179-0.0291
88-0.0800-0.1004-0.0445
89-0.0208-0.0452-0.0407
900.03010.03390.0866
910.03620.0235-0.0530
92-0.0639-0.0591-0.0389
930.0182-0.0668-0.0511
94-0.0752-0.0788-0.0325
95-0.02840.00720.0563
96-0.0865-0.0878-0.0403
97-0.0206-0.0415-0.0530
980.02360.0499-0.0476
990.03090.04860.0444
100-0.0352-0.1167-0.0508
101-0.0472-0.1046-0.0425
102-0.02140.0112-0.0503
103-0.0419-0.1056-0.0744
104-0.0248-0.06280.0228
105-0.0342-0.06520.0642
1060.01790.00080.0935
1070.03130.01700.0992
108-0.0489-0.0394-0.0335
1090.08880.07460.0976
110-0.0243-0.0209-0.0146
1110.04650.10510.0807
112-0.0112-0.01930.0089
113-0.0224-0.01560.0119
1140.07960.03840.0412
115-0.0571-0.0505-0.0319
116-0.0196-0.0259-0.0808
117-0.0354-0.0783-0.0735
118-0.0165-0.0518-0.0396
1190.10800.21020.1157
120-0.0106-0.0352-0.0468
1210.00730.06940.1346
1220.05780.03500.1277
1230.04320.00630.0270
1240.03200.03730.0742
1250.02940.0193-0.0240
1260.05470.1735-0.0544
127-0.00420.0054-0.0388
1280.0663-0.00450.0346
1290.02050.0434-0.0345
1300.0179-0.01170.0132
1310.0277-0.0101-0.0394
132-0.0943-0.0654-0.0477
133-0.1169-0.0825-0.0445
1340.09500.12400.1037
1350.09090.08570.0804
136-0.1107-0.0809-0.0482
1370.0299-0.00570.0721
1380.06830.05200.0935
139-0.0625-0.0077-0.0461
1400.00280.05050.0249
141-0.0779-0.0576-0.0446
142-0.0543-0.0644-0.0389
1430.01020.03140.0942
144-0.00230.01850.1373
145-0.0489-0.0332-0.0158
1460.07280.0641-0.0053
1470.08680.06310.0853
148-0.0789-0.0594-0.0497
1490.03080.01800.0319
150-0.05020.05050.0179
1510.20370.16400.3110
152-0.0930-0.0873-0.0499
1530.03830.11350.0898
1540.04820.06040.1166
1550.05060.04100.1113
156-0.00630.0068-0.0255
1570.02870.03270.0014
158-0.0820-0.0930-0.0561
1590.05910.08940.1351
1600.07690.05810.0298
161-0.0813-0.0583-0.0323
1620.03260.09460.0723
163-0.0086-0.03090.0279
164-0.0038-0.0053-0.0396
165-0.0523-0.0615-0.0479
166-0.1579-0.1041-0.0524
1670.06320.03470.1149
1680.03310.05620.0145
1690.12850.08040.0505
1700.0006-0.04390.0297
171-0.0820-0.0704-0.0396
1720.10870.15350.1432
173-0.0711-0.0557-0.0528
174-0.0991-0.1004-0.0462
175-0.0571-0.0402-0.0228
1760.05260.0699-0.0341
1770.0042-0.04910.0155
1780.01400.0269-0.0161
179-0.0109-0.03420.0028
1800.16130.15450.1247
1810.0306-0.0131-0.0398
182-0.0537-0.0502-0.0257
1830.0045-0.0457-0.0040
184-0.0465-0.0139-0.0457
1850.13050.03890.0091
186-0.0448-0.06100.0488
187-0.0154-0.0068-0.0512
1880.0337-0.0151-0.0423
1890.08060.01480.0595
190-0.0273-0.01140.0093
1910.02750.04780.0249
192-0.0306-0.0250-0.0011
193-0.0384-0.0230-0.0474
194-0.1442-0.1062-0.0584
1950.03450.0515-0.0409
1960.02070.0272-0.0437
1970.06460.0168-0.0630
1980.0029-0.02840.0105
1990.04850.12980.0481
200-0.0218-0.0757-0.0520
2010.01700.04360.0671
202-0.1346-0.1472-0.0936
203-0.01550.0197-0.0399
2040.19140.20920.1648
205-0.0475-0.0909-0.0783
206-0.0550-0.0357-0.0464
207-0.0636-0.0187-0.0498
208-0.0595-0.0683-0.0467
2090.03490.02300.0180
2100.0574-0.0293-0.0264
211-0.0420-0.0618-0.0346
2120.05190.01850.0312
2130.09230.03420.0770
214-0.07830.0238-0.0567
2150.0120-0.0112-0.0201
216-0.0294-0.00630.0044
2170.05130.0434-0.0360
218-0.0796-0.0930-0.0522
219-0.0581-0.0515-0.0586
2200.05400.03810.1262
221-0.0019-0.0794-0.0391
2220.0055-0.04780.0716
223-0.0574-0.0752-0.0361
2240.02480.01770.0932
2250.0068-0.0016-0.0374
2260.06870.00520.0089
2270.08680.09570.0705
228-0.0537-0.0046-0.0729
2290.01090.0604-0.0490
230-0.0461-0.0373-0.0458
2310.0060-0.0115-0.0072
232-0.1066-0.1104-0.0618
2330.0296-0.00440.0020
234-0.0721-0.0229-0.0382
235-0.1278-0.0158-0.0370
236-0.0156-0.01290.0071
237-0.0454-0.0091-0.0459
2380.0543-0.0049-0.0494
239-0.0451-0.0704-0.0465
240-0.0660-0.0604-0.0459
2410.0321-0.0271-0.0356
242-0.1326-0.1346-0.0663
243-0.04720.0234-0.0561
2440.07010.07940.0129
245-0.0220-0.0452-0.0428
246-0.0001-0.0694-0.0364
247-0.0340-0.0470-0.0379
2480.0006-0.0276-0.0515
2490.01910.01480.0446
Source: M2_RECENT_DPO_LORA_SPEAKER_EMOTION_STUDY_2026-10-02.html · SHA-256 51c90e04b8d0c373…
g02GermanDPO acting study

HVSAC-014 · de · targets: Anger, Sadness, Affection

Generated TTS prediction · Rank 64 × 1.00 · C01 · 17.1 s

Du darfst mir das nicht auch noch nehmen! Ich habe den ganzen Tag versucht, ruhig zu bleiben, doch die Wellen sind lauter als meine Ausreden... ich bin wütend, ich bin verzweifelt, und ich liebe dich noch immer. Dann soll das Meer es eben hören.

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Fatigue Exhaustion
z +3.63 / +3.31raw 1.98 / 1.81
Sadness
z +2.61 / +2.31raw 1.98 / 1.75
Disappointment
z +2.56 / +2.59raw 1.79 / 1.82
Bitterness
z +2.42 / +2.49raw 1.02 / 1.05
Longing
z +2.38 / +2.05raw 1.59 / 1.38
Shame
z +2.23 / +1.83raw 1.15 / 0.95
Helplessness
z +2.13 / +1.96raw 1.48 / 1.36

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Tension (TENS)
z +1.52 / +1.68raw 2.86 / 2.99
Ranting or angry style (S_RANT)
z +1.26 / +1.32raw 2.84 / 2.91
Perceived voice age (AGEV)
z +1.22 / +1.03raw 4.49 / 4.26
Chest resonance (R_CHST)
z +1.21 / +1.19raw 3.16 / 3.14
Narrator style (S_NARR)
z +0.98 / +1.05raw 2.73 / 2.83
Cartoonish style (S_CART)
z +0.70 / +0.98raw 2.55 / 2.94
Storytelling style (S_STRY)
z +0.66 / +0.98raw 3.39 / 3.82

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z -0.64 / -0.60raw 1.29 / 1.36
Vocal-burst blend (0–10)z -2.04 / -2.00raw 1.10 / 1.19
Content enjoyment (CE)z +0.40 / +0.58raw 5.21 / 5.39
Production quality (PQ)z +0.80 / +0.80raw 7.75 / 7.74
Speech signal quality, calibratedz +0.30 / +0.29raw 3.64 / 3.63
Background quality, calibratedz +0.73 / +0.63raw 4.06 / 3.98
Overall quality, calibratedz +0.82 / +0.76raw 3.34 / 3.30
P.808 MOS estimatez +1.14 / +0.98raw 3.99 / 3.91
score content enjoymentz +1.09 / +1.12raw 4.84 / 4.85
score overall qualityz +0.95 / +0.68raw 3.13 / 2.99
Background Noisez -1.25 / -0.61raw 0.48 / 0.65
Tempo (TEMP)z -1.20 / -0.80raw 0.69 / 1.16
Arousal (AROU)z -0.54 / -0.14raw 1.74 / 2.20
Valence (VALN)z -0.50 / -0.38raw 1.76 / 1.90
Recording quality (RCQL)z -0.64 / -0.58raw 2.18 / 2.23
Background noise (BKGN)z -0.03 / +0.12raw 2.68 / 2.79
talking speedz +0.27 / -0.37raw 10.05 / 7.63
Characters per second (CPS): 14.36 raw · z +1.42

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8430.857
Identity · 250D0.7910.780

Vocal-burst regions

Tiny
Base
0 s17.1 s

Verified burst intervals are unavailable for this clip.

Tiny: 1 regions, count-head estimate 0.98; Base: 1 regions, count-head estimate 0.88.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
Tiny12.723.380.66
Base12.723.380.66
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.760+1.320.685+1.190.0070.570
1Amusementemo_Amusement0.251+0.280.357+0.40-0.0010.885
2Angeremo_Anger0.793+1.291.004+1.630.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise-0.007-0.02-0.016-0.040.0050.572
4Aweemo_Awe-0.001-0.00-0.020-0.050.0000.427
5Bitternessemo_Bitterness1.021+2.421.046+2.490.0130.415
6Concentrationemo_Concentration0.378-0.350.262-0.500.6340.741
7Confusionemo_Confusion0.707+1.310.482+0.890.0010.540
8Contemplationemo_Contemplation0.913+1.390.745+1.110.0820.598
9Contemptemo_Contempt0.085+0.180.300+0.65-0.0000.464
10Contentmentemo_Contentment0.010+0.020.045+0.090.0000.514
11Disappointmentemo_Disappointment1.788+2.561.816+2.590.0000.700
12Disgustemo_Disgust0.192+0.380.416+0.840.0070.489
13Distressemo_Distress1.621+1.971.532+1.860.0000.821
14Doubtemo_Doubt0.504+1.070.487+1.030.0060.466
15Elationemo_Elation-0.104-0.16-0.034-0.060.0120.728
16Embarrassmentemo_Embarrassment0.555+1.190.628+1.360.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.819+0.890.532+0.580.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion1.981+3.631.812+3.310.0280.539
19Fearemo_Fear0.258+0.400.054+0.060.0150.610
20Helplessnessemo_Helplessness1.476+2.131.360+1.96-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.009-0.140.083-0.050.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability1.220+1.521.599+2.060.1490.705
23Infatuationemo_Infatuation0.552+1.060.253+0.50-0.0150.533
24Interestemo_Interest1.135-0.500.999-0.651.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness0.616+1.150.477+0.90-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.342+0.640.294+0.530.0590.442
27Longingemo_Longing1.593+2.381.384+2.050.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.162+0.200.306+0.470.0550.534
29Painemo_Pain0.745+1.290.659+1.14-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.020+0.010.006-0.020.0170.614
31Prideemo_Pride0.005+0.01-0.019-0.040.0020.509
32Reliefemo_Relief0.405+0.550.587+0.810.0330.681
33Sadnessemo_Sadness1.981+2.611.755+2.310.0000.759
34Sexual Lustemo_Sexual_Lust0.172+0.290.118+0.180.0300.494
35Shameemo_Shame1.149+2.230.948+1.830.0130.510
36Sournessemo_Sourness0.543+1.250.796+1.82-0.0050.439
37Teasingemo_Teasing0.665+1.221.153+2.110.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.238+0.280.337+0.420.0410.697
39Triumphemo_Triumph-0.024-0.03-0.044-0.07-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg4.493+1.224.256+1.032.9841.238
41Arousal (AROU)vn_AROU_reg1.737-0.542.200-0.142.3591.147
42Arousal shift (ARSH)vn_ARSH_reg1.027-1.791.109-1.702.5800.867
43Vocal attack (ATCK)vn_ATCK_reg2.741+0.452.865+0.542.1621.295
44Background noise (BKGN)vn_BKGN_reg2.679-0.032.792+0.122.7030.761
45Brightness (BRGT)vn_BRGT_reg1.606-0.471.723-0.331.9910.813
46Speech chunking (CHNK)vn_CHNK_reg0.981-0.701.113-0.591.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg1.960-0.062.001-0.022.0180.945
48Cognitive load (COGL)vn_COGL_reg2.433+0.322.406+0.292.0941.061
49Dynamic arc (DARC)vn_DARC_reg1.232-0.281.625+0.011.6151.380
50Disfluency (DFLU)vn_DFLU_reg3.253+0.653.258+0.652.4881.184
51Emphasis (EMPH)vn_EMPH_reg3.025+0.273.595+0.652.6071.524
52Esthetics (ESTH)vn_ESTH_reg1.799-0.772.124-0.362.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg-0.079-0.510.098+0.040.0860.322
54Focus (FOCS)vn_FOCS_reg1.452-0.811.578-0.702.3441.102
55Fullness (FULL)vn_FULL_reg1.174-0.681.488-0.331.7780.889
56Perceived gender (GEND)vn_GEND_reg5.424+0.865.424+0.864.0551.588
57Harmonicity (HARM)vn_HARM_reg2.202-0.332.093-0.462.4800.836
58Metallic character (METL)vn_METL_reg1.998+0.391.994+0.391.6210.958
59Pitch range (RANG)vn_RANG_reg3.490+0.643.665+0.842.9140.897
60Recording quality (RCQL)vn_RCQL_reg2.182-0.642.235-0.582.7540.888
61Vocal register (REGS)vn_REGS_reg-0.226-1.12-0.079-1.011.1981.268
62Respiration (RESP)vn_RESP_reg2.312-0.352.639-0.042.6801.056
63Roughness (ROUG)vn_ROUG_reg1.897-0.152.157+0.142.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg3.156+1.213.136+1.192.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg1.500-0.791.606-0.662.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg1.542-0.361.653-0.251.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg0.778-0.781.208-0.271.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg3.110+0.303.431+0.612.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg1.803-0.511.767-0.562.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg2.395+0.402.423+0.442.0700.803
71Smoothness (SMTH)vn_SMTH_reg1.623-0.481.768-0.342.1431.093
72Stance (STNC)vn_STNC_reg2.876+0.213.123+0.432.6331.137
73Speech structure (STRU)vn_STRU_reg2.112-0.032.141+0.002.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg0.650-0.510.413-0.681.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg1.455+0.351.337+0.241.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg2.546+0.702.936+0.981.5601.401
77Casual style (S_CASU)vn_S_CASU_reg2.664-0.883.008-0.613.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg1.990-0.462.660+0.052.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg2.978+0.223.442+0.632.7211.148
80Formal style (S_FORM)vn_S_FORM_reg0.958-0.230.753-0.431.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg3.481+0.513.586+0.592.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg2.725+0.982.832+1.051.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.718+0.550.492+0.150.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg2.334-0.012.583+0.182.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg2.840+1.262.910+1.321.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg3.387+0.663.823+0.982.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg-0.475-0.91-0.366-0.830.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg2.081+0.302.062+0.291.5971.618
89Tempo (TEMP)vn_TEMP_reg0.689-1.201.163-0.802.1171.190
90Tension (TENS)vn_TENS_reg2.856+1.522.986+1.681.5590.852
91Valence (VALN)vn_VALN_reg1.765-0.501.899-0.382.3281.118
92Valence shift (VALS)vn_VALS_reg0.885-1.511.065-1.352.5181.079
93Velocity flux (VFLX)vn_VFLX_reg-1.196-2.86-1.121-2.792.2561.208
94Volatility (VOLT)vn_VOLT_reg2.523-0.072.874+0.332.5860.886
95Vulnerability (VULN)vn_VULN_reg2.609-0.042.847+0.202.6480.984
96Warmth (WARM)vn_WARM_reg2.393-0.202.547-0.032.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_61.294-0.641.363-0.602.3321.621
98Vocal-burst blend (0–10)blend_0_101.102-2.041.187-2.006.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.340-2.240.355-2.191.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.376+0.433.170-0.273.2500.293
101Arousaleiv_extra:Arousal1.460+0.391.690+0.921.2890.435
102Authenticityeiv_extra:Authenticity2.927+1.702.878+1.382.6670.153
103Background Noiseeiv_extra:Background_Noise0.476-1.250.645-0.610.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.423+1.271.439+1.321.0400.303
105Gendereiv_extra:Gender-0.685-0.06-0.604+0.02-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.861+0.131.889+0.251.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive1.948+0.192.157+0.631.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.188+0.132.119-0.082.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous0.934+0.011.109+0.530.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.324+0.140.417+0.430.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.606+0.710.434+0.180.3760.324
112Valenceeiv_extra:Valence-0.606-1.61-0.668-1.770.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.304-0.941.211-1.151.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold-0.184-0.78-0.106-0.520.0510.300
115score background qualityeiv_extra:score_background_quality3.716+0.603.654+0.513.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.838+1.094.847+1.124.5070.303
117score overall qualityeiv_extra:score_overall_quality3.131+0.952.994+0.682.6520.507
118score speech qualityeiv_extra:score_speech_quality1.912+0.721.876+0.471.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.208+0.405.391+0.584.7991.018
120Content usefulness (CU)audiobox:CU6.335+0.696.077+0.435.6491.001
121Production complexity (PC)audiobox:PC1.707-0.391.915-0.262.3311.617
122Production quality (PQ)audiobox:PQ7.751+0.807.744+0.807.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw4.163+0.394.137+0.353.9000.669
124Background quality, rawdnsmos:BAK_raw4.307+0.984.245+0.923.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.865+0.963.808+0.893.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.640+0.303.632+0.293.4880.501
127Background quality, calibrateddnsmos:BAK4.064+0.733.977+0.633.3890.928
128Overall quality, calibrateddnsmos:OVRL3.340+0.823.300+0.762.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.989+1.143.907+0.983.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.683+1.050.630+0.970.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.227-1.300.379-1.030.9470.554
132Agevoiceclap_attribute:Age3.391+0.303.306+0.143.2290.547
133Amusementvoiceclap_attribute:Amusement0.276-0.200.357-0.060.3920.572
134Angervoiceclap_attribute:Anger1.620+1.241.623+1.250.9920.506
135Arousalvoiceclap_attribute:Arousal1.578+0.251.725+0.491.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.446-0.130.734+0.470.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.865+0.572.815+0.272.7700.167
138Awevoiceclap_attribute:Awe0.168-1.430.257-1.090.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.431-0.520.599+0.120.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.952+1.520.934+1.450.5920.236
141Concentrationvoiceclap_attribute:Concentration0.993-0.420.981-0.441.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.331+0.561.562+1.051.0690.468
143Confusionvoiceclap_attribute:Confusion1.040+1.191.114+1.410.6430.334
144Contemplationvoiceclap_attribute:Contemplation1.084+0.750.974+0.480.7800.406
145Contemptvoiceclap_attribute:Contempt0.876+0.490.850+0.420.7090.339
146Contentmentvoiceclap_attribute:Contentment0.206-1.250.295-0.960.5880.307
147Disappointmentvoiceclap_attribute:Disappointment1.469+2.431.419+2.260.7600.291
148Disgustvoiceclap_attribute:Disgust0.742+0.570.799+0.840.6250.207
149Distressvoiceclap_attribute:Distress1.345+1.191.278+1.000.9270.350
150Doubtvoiceclap_attribute:Doubt0.873+1.800.844+1.640.5350.188
151Elationvoiceclap_attribute:Elation0.263-0.700.248-0.730.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.447+0.440.503+0.620.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.852+0.710.681+0.060.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion1.220+2.901.033+2.190.4620.261
155Fearvoiceclap_attribute:Fear0.854+1.110.853+1.110.6180.212
156Gendervoiceclap_attribute:Gender-0.653-0.50-0.579-0.43-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness1.131+1.761.084+1.590.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched2.104+0.702.047+0.551.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.336-0.930.315-0.990.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability1.849+1.471.857+1.491.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.421-0.560.459-0.450.6080.332
162Interestvoiceclap_attribute:Interest0.672-2.111.077-1.181.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.959+0.990.898+0.850.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.721+0.810.750+0.930.5230.245
165Longingvoiceclap_attribute:Longing0.741+0.920.657+0.540.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.839+0.020.905+0.210.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive1.776+0.001.935+0.271.7750.594
168Painvoiceclap_attribute:Pain0.697+1.440.761+1.740.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.273-0.770.395-0.360.5030.301
170Pridevoiceclap_attribute:Pride1.172+1.121.069+0.830.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.157-0.482.000-0.962.3130.328
172Reliefvoiceclap_attribute:Relief1.099+0.181.333+0.741.0250.415
173Sadnessvoiceclap_attribute:Sadness1.244+1.571.057+0.980.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.659-0.250.674-0.230.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.190-1.060.193-1.060.6350.419
176Shamevoiceclap_attribute:Shame1.037+1.800.925+1.410.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh-0.034-0.540.060-0.280.1600.360
178Sournessvoiceclap_attribute:Sourness1.075+1.271.152+1.540.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.854+0.640.711+0.250.6230.359
180Teasingvoiceclap_attribute:Teasing0.817+0.731.008+1.230.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude0.868-0.970.916-0.881.3840.529
182Triumphvoiceclap_attribute:Triumph0.890+0.480.867+0.420.7160.359
183Valencevoiceclap_attribute:Valence-1.344-2.11-1.314-2.06-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached1.432-0.371.201-0.791.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold-0.562-1.24-0.737-1.68-0.0660.400
186durationvoiceclap_attribute:duration16.014+0.8018.563+1.2010.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.729+0.973.869+1.253.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.811+1.054.818+1.104.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality3.068+1.053.164+1.282.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.938+0.681.911+0.291.8910.069
191talking speedvoiceclap_attribute:talking_speed10.053+0.277.627-0.379.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.08310.03850.0136
1-0.0389-0.0069-0.0754
2-0.0237-0.0492-0.0744
3-0.1192-0.1376-0.0769
40.14100.13970.0614
50.08920.13690.1392
60.13050.19530.1639
7-0.0326-0.0269-0.0695
80.0073-0.0075-0.0510
9-0.0461-0.1034-0.1142
100.07880.10050.0439
11-0.0692-0.0250-0.0510
12-0.01740.03460.0361
13-0.0640-0.0663-0.0219
140.0119-0.0495-0.0790
15-0.0849-0.0809-0.0680
16-0.03390.0050-0.0228
17-0.0300-0.0485-0.0789
180.04130.07730.0827
19-0.0205-0.0144-0.0669
20-0.0193-0.00850.0935
21-0.00490.0681-0.0002
22-0.0335-0.0305-0.0228
23-0.0045-0.0040-0.0443
24-0.0648-0.0667-0.0612
25-0.0775-0.0816-0.0667
260.00840.0125-0.0712
27-0.0762-0.0506-0.0707
28-0.0849-0.0788-0.0836
290.01290.0118-0.0812
300.18620.14610.1342
310.11750.11260.1227
320.06960.02280.0162
330.02460.0577-0.0836
34-0.01480.00040.0368
350.05740.09270.1516
36-0.0322-0.04050.0242
370.13320.19110.1985
38-0.1201-0.1262-0.0757
39-0.0605-0.0745-0.0765
40-0.1044-0.1119-0.0732
41-0.0065-0.0517-0.0961
42-0.0090-0.0499-0.0280
43-0.0988-0.0912-0.0781
44-0.0298-0.0143-0.0165
450.0285-0.0189-0.1020
460.0128-0.0285-0.0459
47-0.1140-0.0984-0.0767
480.03130.05100.0377
49-0.0762-0.0720-0.0688
500.20540.29230.2622
510.0302-0.01130.0840
52-0.0648-0.0510-0.0393
53-0.0936-0.0838-0.0773
540.06270.07770.1101
55-0.0596-0.0244-0.0368
560.08960.08630.0864
57-0.0683-0.0372-0.0745
58-0.1210-0.1283-0.0808
590.03480.05170.0561
600.06570.04440.1104
61-0.0796-0.0592-0.0699
62-0.0779-0.1133-0.0709
63-0.0044-0.0216-0.0575
640.05160.04850.0003
65-0.0783-0.0699-0.0619
660.03590.0331-0.0715
670.09970.10760.1566
680.02810.0081-0.0802
69-0.0444-0.0524-0.0695
70-0.1053-0.0998-0.0786
710.07400.07910.0314
72-0.0322-0.0841-0.0734
730.01690.0927-0.0015
74-0.0923-0.1041-0.0829
750.02330.0741-0.0714
760.15320.12400.1016
770.13750.12330.1556
780.07830.02850.0999
79-0.0722-0.0777-0.0658
800.08010.06170.0833
810.12790.09270.0700
820.14190.12480.1306
83-0.0161-0.0435-0.0597
84-0.1358-0.1084-0.0941
85-0.00470.07700.1128
86-0.0261-0.0237-0.0879
87-0.0468-0.0428-0.0767
880.0079-0.00390.0593
890.05700.05310.0353
900.0011-0.0659-0.0434
910.15930.21810.1021
92-0.1297-0.1183-0.0737
93-0.0190-0.0032-0.0072
940.00920.04460.0452
950.03850.01660.1127
96-0.01350.0098-0.0386
970.0289-0.03420.0269
98-0.0831-0.0773-0.0767
990.36550.24380.2898
1000.13840.11260.1153
101-0.00610.0481-0.0466
102-0.1506-0.0542-0.0812
103-0.0335-0.0606-0.0801
104-0.0775-0.0517-0.0742
105-0.1262-0.0773-0.0732
106-0.1532-0.1533-0.0912
1070.15320.12550.0688
1080.09360.09550.0872
109-0.0622-0.0563-0.0744
1100.09010.10190.0890
111-0.0479-0.0681-0.0880
112-0.0640-0.0610-0.0748
113-0.0753-0.0351-0.0829
1140.0289-0.0027-0.0599
115-0.0313-0.0431-0.0719
116-0.0487-0.0941-0.0650
117-0.0849-0.1084-0.0883
118-0.0379-0.0367-0.0739
1190.02700.0731-0.0095
1200.25940.22670.2325
121-0.1044-0.0962-0.0524
122-0.1031-0.0749-0.0779
123-0.0313-0.0962-0.0876
124-0.0403-0.0060-0.0798
125-0.0788-0.0873-0.0756
126-0.1040-0.0977-0.0654
1270.08790.10190.0667
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.0677-0.0851-0.0443
1-0.0515-0.0494-0.0389
2-0.1192-0.0290-0.0343
3-0.0430-0.0814-0.0764
4-0.03850.0143-0.0366
50.04430.03710.0656
60.05960.06520.1120
70.11770.07720.1217
80.0461-0.05240.0302
90.02730.0025-0.0032
10-0.1076-0.0298-0.0291
11-0.0459-0.0204-0.0461
120.08010.10920.1447
130.06500.11230.0652
14-0.0385-0.0584-0.0046
150.09330.0486-0.0474
160.06850.10470.0520
17-0.02860.03260.1077
180.0052-0.0185-0.0409
190.01570.04800.0567
20-0.01650.0214-0.0403
21-0.0557-0.0505-0.0373
22-0.0250-0.0516-0.0417
23-0.0298-0.0348-0.0389
24-0.00890.0187-0.0305
25-0.0639-0.1085-0.0434
260.17490.15600.1337
27-0.0565-0.0350-0.0327
28-0.0164-0.00040.0222
29-0.0397-0.0482-0.0387
30-0.0281-0.0201-0.0350
31-0.0979-0.0844-0.0505
320.03440.00680.0047
330.04760.05690.0986
340.08130.12210.0589
35-0.03310.0231-0.0332
360.03190.0723-0.0274
37-0.0290-0.0614-0.0278
380.0207-0.00320.0810
390.01860.01610.1051
40-0.0453-0.0554-0.0466
41-0.0096-0.00690.0515
420.01830.02300.0294
430.10680.03670.0697
44-0.0492-0.0135-0.0385
45-0.0466-0.0727-0.0461
46-0.0211-0.0196-0.0591
470.04040.09790.0998
48-0.0000-0.01910.0148
49-0.0492-0.0087-0.0309
500.03040.0350-0.0307
51-0.04900.0046-0.0163
52-0.0774-0.0667-0.0397
530.04370.05610.0352
54-0.0018-0.09950.0625
550.04330.06780.0930
56-0.0002-0.0226-0.0410
570.18580.13490.1589
58-0.0009-0.02920.0328
590.0003-0.0017-0.0287
600.0168-0.04330.0258
61-0.0211-0.0109-0.0374
62-0.0507-0.0114-0.0037
630.0076-0.0235-0.0061
640.01250.0207-0.0040
650.05570.03370.0889
660.02390.10550.0355
670.14170.09420.1766
68-0.0201-0.0173-0.0433
690.0134-0.0174-0.0236
70-0.0116-0.01180.0573
710.00020.0213-0.0421
720.01850.0193-0.0156
730.04680.09870.0970
74-0.0523-0.0501-0.0434
750.10990.06330.1294
760.06770.00990.0788
770.0201-0.0352-0.0190
780.12230.1175-0.0006
790.01040.0192-0.0343
80-0.02690.0183-0.0427
81-0.0381-0.0550-0.0482
82-0.0592-0.0377-0.0386
83-0.0635-0.0201-0.0357
84-0.0435-0.0625-0.0605
85-0.0339-0.0712-0.0189
860.02550.02020.0145
87-0.00920.01920.0212
88-0.0162-0.00690.0021
89-0.0370-0.0806-0.0388
900.06850.09490.0958
910.05610.08660.0466
92-0.0455-0.0591-0.0371
93-0.0561-0.0870-0.0487
94-0.0248-0.0204-0.0558
95-0.0219-0.0155-0.0448
96-0.0863-0.0573-0.0384
97-0.0627-0.0953-0.0506
98-0.02320.0089-0.0066
99-0.0121-0.0527-0.0537
100-0.0065-0.0160-0.0202
101-0.0348-0.0367-0.0038
102-0.0350-0.0761-0.0480
103-0.0490-0.1062-0.0139
104-0.0261-0.0284-0.0193
105-0.0352-0.0284-0.0225
106-0.0466-0.0339-0.0331
1070.03970.04350.0414
1080.10840.02510.0390
1090.11150.04690.1041
1100.08940.03900.1049
1110.12310.09080.1195
112-0.0090-0.00560.0964
1130.0197-0.0065-0.0268
114-0.0315-0.0516-0.0599
115-0.0375-0.0150-0.0304
116-0.03250.0105-0.0474
117-0.1022-0.0787-0.0701
118-0.00940.0039-0.0408
119-0.0712-0.0949-0.0442
120-0.0685-0.0535-0.0547
1210.08590.09610.1307
122-0.0306-0.01750.0160
123-0.0319-0.0482-0.0444
124-0.0596-0.0469-0.0428
125-0.0171-0.0294-0.0417
1260.10300.05050.0227
127-0.0263-0.0397-0.0442
128-0.0490-0.0761-0.0453
129-0.0166-0.0512-0.0249
130-0.0214-0.0490-0.0432
131-0.0673-0.0652-0.0376
132-0.0416-0.0904-0.0455
133-0.0751-0.0443-0.0424
1340.09480.11150.0859
135-0.0048-0.0360-0.0432
136-0.0677-0.0254-0.0460
137-0.0898-0.0471-0.0399
138-0.00100.03010.0784
1390.0230-0.0019-0.0437
140-0.0573-0.0795-0.0506
1410.13620.06440.0473
142-0.0298-0.0262-0.0371
1430.02880.10170.0609
144-0.01940.01310.0560
145-0.0441-0.0441-0.0101
1460.03210.04160.0028
1470.05810.03150.0153
1480.06540.05050.0974
1490.01890.05690.0687
1500.06460.09340.0055
1510.12850.09790.1863
1520.0210-0.0204-0.0476
153-0.0099-0.0007-0.0676
1540.08050.09420.0908
1550.05770.09150.0113
156-0.02170.0089-0.0430
1570.11840.09870.0790
1580.0101-0.0433-0.0301
159-0.0304-0.0303-0.0503
160-0.0302-0.0103-0.0170
161-0.0894-0.0697-0.0280
162-0.0708-0.0439-0.0498
1630.02540.05580.0634
1640.04040.01270.0010
165-0.0171-0.0671-0.0592
1660.04840.07910.0728
1670.04260.06710.0359
168-0.1061-0.0648-0.0607
1690.04760.07460.0390
170-0.0488-0.0686-0.0402
1710.04220.0326-0.0378
172-0.0542-0.0048-0.0093
1730.06970.00940.0803
174-0.0689-0.0494-0.0441
175-0.0542-0.0384-0.0411
1760.19660.14990.2122
177-0.0515-0.0337-0.0354
1780.09100.13560.1599
1790.08090.08930.0790
1800.06810.16730.1648
181-0.0809-0.1040-0.0612
182-0.0604-0.0712-0.0371
183-0.0250-0.0689-0.0193
184-0.0220-0.0097-0.0436
185-0.0087-0.0028-0.0284
1860.08130.04650.0494
187-0.0786-0.0294-0.0489
1880.0379-0.00450.0318
1890.00690.08550.0861
1900.09250.07420.1501
1910.0298-0.00680.0005
1920.02920.08930.0772
193-0.0724-0.0701-0.0452
1940.05110.10620.0577
1950.03640.04030.0275
1960.15870.17480.2159
197-0.0577-0.0199-0.0601
198-0.0387-0.0135-0.0415
1990.02330.03130.0304
200-0.0790-0.1085-0.0496
2010.08940.04730.0030
202-0.1564-0.0753-0.0893
2030.08900.0595-0.0131
2040.08520.08820.1144
205-0.1301-0.1529-0.0747
206-0.0286-0.0137-0.0443
207-0.0530-0.0863-0.0529
208-0.0503-0.0591-0.0445
209-0.0352-0.0569-0.0461
2100.13160.12810.0832
211-0.02500.01380.0295
2120.10370.05990.0808
2130.21370.17400.1219
2140.09910.0414-0.0060
2150.02790.00490.0102
2160.08010.11150.0733
217-0.0809-0.1017-0.0786
218-0.1006-0.0390-0.0498
219-0.0263-0.0478-0.0381
2200.04080.00210.0326
2210.04220.02790.1632
222-0.0341-0.0168-0.0499
223-0.0499-0.0610-0.0344
2240.0001-0.0224-0.0392
2250.02280.0022-0.0150
2260.04800.03090.0265
227-0.0375-0.0882-0.0567
228-0.0581-0.0046-0.0236
2290.10370.14920.1219
230-0.0666-0.0211-0.0437
231-0.02180.0085-0.0009
2320.02130.0092-0.0161
233-0.0732-0.0403-0.0722
2340.06120.07120.0298
235-0.00170.0142-0.0353
236-0.0414-0.0335-0.0369
237-0.0197-0.0172-0.0195
2380.03120.0337-0.0471
239-0.0202-0.0428-0.0444
240-0.0542-0.0114-0.0438
2410.0219-0.03350.0669
242-0.0247-0.0663-0.0632
2430.0053-0.01590.0007
244-0.03270.0058-0.0456
245-0.0685-0.1386-0.0408
246-0.0530-0.0674-0.0348
247-0.0565-0.0772-0.0361
248-0.0588-0.0697-0.0491
249-0.0375-0.0006-0.0205
Source: M2_RECENT_DPO_LORA_SPEAKER_EMOTION_STUDY_2026-10-02.html · SHA-256 4d40ef1f8b7810a1…
g03EnglishDPO acting study

HVSAC-017 · en · targets: Pain, Fear, Triumph

Generated TTS prediction · Rank 16 × 1.00 · C01 · 16.8 s

My fingers have gone numb, and the station lights keep blinking. I can hear the wind pushing at the door... please tell me the heater is still working. I need to stay awake until the crew gets here.

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Fatigue Exhaustion
z +3.99 / +3.12raw 2.18 / 1.71
Helplessness
z +3.47 / +3.17raw 2.41 / 2.20
Distress
z +3.33 / +3.06raw 2.73 / 2.52
Fear
z +3.14 / +3.25raw 1.93 / 1.99
Sadness
z +3.08 / +2.71raw 2.33 / 2.06
Disappointment
z +2.81 / +2.65raw 1.97 / 1.85
Shame
z +2.47 / +1.84raw 1.28 / 0.95

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Tension (TENS)
z +1.78 / +2.13raw 3.07 / 3.38
Newsreader style (S_NEWS)
z +1.11 / +1.39raw 1.04 / 1.20
Mixed resonance (R_MIXD)
z +0.97 / +1.27raw 2.25 / 2.51
Whispered talking style (S_WHIS)
z +1.15 / +0.99raw 3.45 / 3.19
Vulnerability (VULN)
z +1.10 / +1.01raw 3.73 / 3.64
Cognitive load (COGL)
z +1.00 / +0.56raw 3.15 / 2.69
Perceived gender (GEND)
z +0.93 / +0.77raw 5.52 / 5.27

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z -0.17 / -0.28raw 2.06 / 1.87
Vocal-burst blend (0–10)z +0.22 / +0.11raw 7.28 / 6.98
Content enjoyment (CE)z +0.42 / +0.99raw 5.22 / 5.80
Production quality (PQ)z +0.64 / +0.28raw 7.61 / 7.29
Speech signal quality, calibratedz -0.47 / -1.45raw 3.25 / 2.76
Background quality, calibratedz +0.46 / -0.04raw 3.81 / 3.35
Overall quality, calibratedz +0.22 / -0.35raw 2.97 / 2.63
P.808 MOS estimatez +0.27 / +0.31raw 3.56 / 3.57
score content enjoymentz -0.25 / +0.21raw 4.43 / 4.57
score overall qualityz +0.52 / +0.38raw 2.91 / 2.84
Background Noisez -0.27 / -0.48raw 0.73 / 0.68
Tempo (TEMP)z -0.98 / -1.14raw 0.95 / 0.76
Arousal (AROU)z -1.24 / -0.77raw 0.94 / 1.47
Valence (VALN)z -1.49 / -1.47raw 0.66 / 0.69
Recording quality (RCQL)z -0.41 / -0.70raw 2.39 / 2.13
Background noise (BKGN)z -0.49 / -0.61raw 2.33 / 2.24
talking speedz +0.75 / -0.21raw 11.86 / 8.23
Characters per second (CPS): 11.77 raw · z +0.73

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8080.876
Identity · 250D0.6990.671

Vocal-burst regions

Tiny
Base
0 s16.8 s

Verified burst intervals are unavailable for this clip.

Tiny: 2 regions, count-head estimate 1.72; Base: 2 regions, count-head estimate 1.80.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
Tiny15.565.720.16
Tiny210.9011.220.32
Base15.565.720.16
Base210.9011.220.32
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.180+0.300.198+0.340.0070.570
1Amusementemo_Amusement0.047+0.050.061+0.07-0.0010.885
2Angeremo_Anger0.094+0.150.021+0.030.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.052+0.080.020+0.030.0050.572
4Aweemo_Awe0.017+0.040.001+0.000.0000.427
5Bitternessemo_Bitterness0.283+0.650.151+0.330.0130.415
6Concentrationemo_Concentration0.887+0.340.887+0.340.6340.741
7Confusionemo_Confusion0.753+1.390.689+1.270.0010.540
8Contemplationemo_Contemplation0.816+1.230.737+1.090.0820.598
9Contemptemo_Contempt0.018+0.04-0.012-0.02-0.0000.464
10Contentmentemo_Contentment0.002+0.00-0.045-0.090.0000.514
11Disappointmentemo_Disappointment1.969+2.811.854+2.650.0000.700
12Disgustemo_Disgust0.012+0.01-0.025-0.070.0070.489
13Distressemo_Distress2.732+3.332.515+3.060.0000.821
14Doubtemo_Doubt0.941+2.010.944+2.020.0060.466
15Elationemo_Elation0.031+0.03-0.003-0.020.0120.728
16Embarrassmentemo_Embarrassment0.639+1.390.627+1.360.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.292+0.310.250+0.270.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion2.176+3.991.711+3.120.0280.539
19Fearemo_Fear1.928+3.141.994+3.250.0150.610
20Helplessnessemo_Helplessness2.411+3.472.201+3.17-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.024-0.12-0.006-0.160.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.462+0.440.379+0.330.1490.705
23Infatuationemo_Infatuation0.160+0.330.120+0.25-0.0150.533
24Interestemo_Interest0.617-1.070.987-0.661.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness0.024+0.090.231+0.46-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.102+0.100.077+0.040.0590.442
27Longingemo_Longing1.647+2.471.518+2.270.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.056+0.000.043-0.020.0550.534
29Painemo_Pain1.232+2.131.066+1.84-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy-0.104-0.200.025+0.010.0170.614
31Prideemo_Pride0.024+0.04-0.009-0.020.0020.509
32Reliefemo_Relief1.052+1.501.079+1.540.0330.681
33Sadnessemo_Sadness2.334+3.082.059+2.710.0000.759
34Sexual Lustemo_Sexual_Lust0.237+0.420.428+0.810.0300.494
35Shameemo_Shame1.276+2.470.952+1.840.0130.510
36Sournessemo_Sourness-0.033-0.06-0.010-0.01-0.0050.439
37Teasingemo_Teasing0.056+0.10-0.031-0.060.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.208+0.240.177+0.200.0410.697
39Triumphemo_Triumph0.004+0.02-0.039-0.06-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg3.380+0.323.191+0.172.9841.238
41Arousal (AROU)vn_AROU_reg0.941-1.241.473-0.772.3591.147
42Arousal shift (ARSH)vn_ARSH_reg0.602-2.280.794-2.062.5800.867
43Vocal attack (ATCK)vn_ATCK_reg0.493-1.290.637-1.182.1621.295
44Background noise (BKGN)vn_BKGN_reg2.330-0.492.235-0.612.7030.761
45Brightness (BRGT)vn_BRGT_reg2.133+0.172.238+0.301.9910.813
46Speech chunking (CHNK)vn_CHNK_reg0.937-0.731.254-0.481.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg1.474-0.581.406-0.652.0180.945
48Cognitive load (COGL)vn_COGL_reg3.150+1.002.693+0.562.0941.061
49Dynamic arc (DARC)vn_DARC_reg-0.367-1.44-0.324-1.411.6151.380
50Disfluency (DFLU)vn_DFLU_reg2.893+0.342.674+0.162.4881.184
51Emphasis (EMPH)vn_EMPH_reg1.331-0.841.607-0.662.6071.524
52Esthetics (ESTH)vn_ESTH_reg2.478+0.102.561+0.202.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg-0.042-0.400.108+0.070.0860.322
54Focus (FOCS)vn_FOCS_reg0.776-1.421.079-1.152.3441.102
55Fullness (FULL)vn_FULL_reg2.462+0.772.548+0.871.7780.889
56Perceived gender (GEND)vn_GEND_reg5.524+0.935.275+0.774.0551.588
57Harmonicity (HARM)vn_HARM_reg2.095-0.462.190-0.352.4800.836
58Metallic character (METL)vn_METL_reg0.771-0.890.846-0.811.6210.958
59Pitch range (RANG)vn_RANG_reg1.673-1.381.930-1.102.9140.897
60Recording quality (RCQL)vn_RCQL_reg2.389-0.412.133-0.702.7540.888
61Vocal register (REGS)vn_REGS_reg0.937-0.210.844-0.281.1981.268
62Respiration (RESP)vn_RESP_reg3.528+0.803.540+0.812.6801.056
63Roughness (ROUG)vn_ROUG_reg1.406-0.711.451-0.662.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg2.598+0.552.471+0.402.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg0.799-1.640.767-1.682.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg1.007-0.900.883-1.021.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg2.249+0.972.507+1.271.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg0.432-2.250.416-2.272.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg1.975-0.282.138-0.062.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg0.382-2.100.596-1.842.0700.803
71Smoothness (SMTH)vn_SMTH_reg1.134-0.921.073-0.982.1431.093
72Stance (STNC)vn_STNC_reg0.910-1.511.355-1.122.6331.137
73Speech structure (STRU)vn_STRU_reg1.662-0.501.484-0.682.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg1.959+0.421.797+0.301.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg0.814-0.261.090+0.001.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg1.373-0.131.326-0.171.5601.401
77Casual style (S_CASU)vn_S_CASU_reg3.190-0.473.489-0.243.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg1.575-0.771.779-0.622.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg1.978-0.652.290-0.382.7211.148
80Formal style (S_FORM)vn_S_FORM_reg1.510+0.311.361+0.161.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg3.007+0.142.777-0.042.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg1.001-0.211.246-0.041.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg1.038+1.111.198+1.390.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg-0.149-1.900.075-1.732.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg0.859-0.270.898-0.241.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg1.335-0.841.705-0.572.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg-0.332-0.80-0.156-0.650.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg3.453+1.153.191+0.991.5971.618
89Tempo (TEMP)vn_TEMP_reg0.952-0.980.764-1.142.1171.190
90Tension (TENS)vn_TENS_reg3.070+1.783.376+2.131.5590.852
91Valence (VALN)vn_VALN_reg0.659-1.490.686-1.472.3281.118
92Valence shift (VALS)vn_VALS_reg0.996-1.410.996-1.412.5181.079
93Velocity flux (VFLX)vn_VFLX_reg-1.083-2.76-0.876-2.592.2561.208
94Volatility (VOLT)vn_VOLT_reg2.877+0.332.863+0.312.5860.886
95Vulnerability (VULN)vn_VULN_reg3.731+1.103.638+1.012.6480.984
96Warmth (WARM)vn_WARM_reg2.797+0.242.677+0.112.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_62.057-0.171.871-0.282.3321.621
98Vocal-burst blend (0–10)blend_0_107.278+0.226.984+0.116.6762.738
99Legacy R_quality (no P3 test labels)R_quality1.063-0.091.008-0.261.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.543+1.003.224-0.093.2500.293
101Arousaleiv_extra:Arousal1.350+0.141.438+0.341.2890.435
102Authenticityeiv_extra:Authenticity2.736+0.452.934+1.752.6670.153
103Background Noiseeiv_extra:Background_Noise0.735-0.270.680-0.480.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.544+1.671.183+0.471.0400.303
105Gendereiv_extra:Gender-0.268+0.35-0.376+0.25-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.833+0.021.806-0.091.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive1.525-0.711.718-0.301.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.066-0.232.163+0.062.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous0.082-2.490.564-1.070.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.453+0.540.481+0.630.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.239-0.420.455+0.250.3760.324
112Valenceeiv_extra:Valence-0.629-1.67-1.056-2.790.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.388-0.751.516-0.461.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold-0.048-0.33-0.153-0.680.0510.300
115score background qualityeiv_extra:score_background_quality3.461+0.233.532+0.333.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.432-0.254.570+0.214.5070.303
117score overall qualityeiv_extra:score_overall_quality2.913+0.522.842+0.382.6520.507
118score speech qualityeiv_extra:score_speech_quality1.738-0.481.798-0.071.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.222+0.425.804+0.994.7991.018
120Content usefulness (CU)audiobox:CU6.526+0.886.167+0.525.6491.001
121Production complexity (PC)audiobox:PC2.555+0.141.719-0.382.3311.617
122Production quality (PQ)audiobox:PQ7.607+0.647.287+0.287.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw3.627-0.412.934-1.443.9000.669
124Background quality, rawdnsmos:BAK_raw3.749+0.483.434+0.203.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.273+0.232.908-0.223.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.251-0.472.761-1.453.4880.501
127Background quality, calibrateddnsmos:BAK3.814+0.463.352-0.043.3890.928
128Overall quality, calibrateddnsmos:OVRL2.974+0.222.627-0.352.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.555+0.273.575+0.313.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p1.001+1.551.028+1.590.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.997+0.091.002+0.100.9470.554
132Agevoiceclap_attribute:Age2.899-0.603.124-0.193.2290.547
133Amusementvoiceclap_attribute:Amusement0.162-0.400.052-0.590.3920.572
134Angervoiceclap_attribute:Anger0.581-0.810.812-0.360.9920.506
135Arousalvoiceclap_attribute:Arousal1.243-0.281.555+0.221.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.350-0.330.547+0.080.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.943+1.042.970+1.192.7700.167
138Awevoiceclap_attribute:Awe0.233-1.180.432-0.410.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.405-0.620.624+0.210.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.608+0.070.795+0.860.5920.236
141Concentrationvoiceclap_attribute:Concentration0.918-0.560.673-1.021.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.160+0.191.125+0.121.0690.468
143Confusionvoiceclap_attribute:Confusion0.726+0.250.752+0.330.6430.334
144Contemplationvoiceclap_attribute:Contemplation1.234+1.120.768-0.030.7800.406
145Contemptvoiceclap_attribute:Contempt0.438-0.800.178-1.570.7090.339
146Contentmentvoiceclap_attribute:Contentment0.500-0.290.110-1.560.5880.307
147Disappointmentvoiceclap_attribute:Disappointment1.072+1.071.344+2.010.7600.291
148Disgustvoiceclap_attribute:Disgust0.363-1.260.520-0.510.6250.207
149Distressvoiceclap_attribute:Distress1.539+1.751.922+2.840.9270.350
150Doubtvoiceclap_attribute:Doubt0.773+1.270.865+1.760.5350.188
151Elationvoiceclap_attribute:Elation-0.274-2.050.003-1.350.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.408+0.320.279-0.090.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness1.092+1.610.978+1.180.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.785+1.240.679+0.830.4620.261
155Fearvoiceclap_attribute:Fear1.059+2.081.087+2.210.6180.212
156Gendervoiceclap_attribute:Gender-0.137+0.01-0.233-0.08-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness1.164+1.881.445+2.930.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.949+0.292.031+0.511.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.144-1.420.159-1.380.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability1.042-0.021.354+0.561.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.570-0.120.417-0.580.6080.332
162Interestvoiceclap_attribute:Interest0.840-1.720.998-1.361.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.679+0.380.455-0.100.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.560+0.150.485-0.150.5230.245
165Longingvoiceclap_attribute:Longing0.691+0.700.782+1.110.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.661-0.500.683-0.430.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive1.427-0.591.668-0.181.7750.594
168Painvoiceclap_attribute:Pain0.877+2.270.953+2.620.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.114-1.290.081-1.400.5030.301
170Pridevoiceclap_attribute:Pride0.762-0.050.527-0.730.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.234-0.242.143-0.522.3130.328
172Reliefvoiceclap_attribute:Relief0.844-0.431.173+0.361.0250.415
173Sadnessvoiceclap_attribute:Sadness1.178+1.361.613+2.750.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.676-0.220.731-0.130.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.376-0.620.007-1.500.6350.419
176Shamevoiceclap_attribute:Shame0.592+0.230.615+0.310.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.415+0.710.396+0.660.1600.360
178Sournessvoiceclap_attribute:Sourness0.531-0.600.362-1.180.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.149-1.320.099-1.460.6230.359
180Teasingvoiceclap_attribute:Teasing0.487-0.140.545+0.010.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.176-0.390.742-1.211.3840.529
182Triumphvoiceclap_attribute:Triumph0.589-0.350.486-0.640.7160.359
183Valencevoiceclap_attribute:Valence-1.324-2.08-1.523-2.39-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached1.496-0.261.474-0.301.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold-0.117-0.13-0.038+0.07-0.0660.400
186durationvoiceclap_attribute:duration17.301+1.0120.394+1.4910.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.425+0.363.342+0.193.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.769+0.774.849+1.314.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.648+0.032.608-0.072.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.929+0.551.894+0.051.8910.069
191talking speedvoiceclap_attribute:talking_speed11.856+0.758.228-0.219.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
0-0.0484-0.1252-0.0913
10.04550.12420.0269
2-0.0704-0.0514-0.0900
3-0.1076-0.0865-0.0930
40.10060.08290.0112
5-0.0353-0.0804-0.0942
60.0195-0.0578-0.0832
7-0.05090.02440.0290
80.0401-0.05750.0068
90.02440.00310.0640
100.0156-0.0031-0.0011
11-0.11530.0008-0.0640
120.01940.0073-0.0288
13-0.0767-0.1160-0.0906
14-0.0353-0.02960.0386
15-0.0100-0.0992-0.0752
16-0.1350-0.0649-0.1136
17-0.01200.03540.0642
18-0.0193-0.0106-0.0363
19-0.03230.0379-0.0334
20-0.0089-0.0267-0.0380
210.10000.13540.1355
22-0.1038-0.0784-0.0863
23-0.1108-0.0743-0.0540
24-0.0923-0.0616-0.0741
25-0.0815-0.0202-0.0780
260.05800.0639-0.0092
27-0.06460.0083-0.0836
28-0.1337-0.1099-0.1011
290.03600.10330.1250
300.26110.19130.1816
310.1497-0.02260.0707
32-0.0666-0.0656-0.0782
330.04620.06770.0702
34-0.1248-0.0555-0.0823
350.19870.20360.2054
36-0.0264-0.0672-0.0986
37-0.0471-0.0285-0.0820
38-0.1465-0.0957-0.0916
39-0.1083-0.0997-0.0926
40-0.1102-0.0794-0.0541
41-0.1102-0.1323-0.1318
42-0.1490-0.0298-0.0926
43-0.0646-0.0132-0.0674
440.14900.13330.1547
45-0.03450.00850.0017
46-0.1108-0.0921-0.0853
47-0.0866-0.0957-0.0928
48-0.0379-0.0415-0.0896
490.0136-0.0992-0.0811
500.02610.09360.0598
51-0.0691-0.0855-0.0917
52-0.0229-0.1064-0.0299
530.04650.0865-0.0070
540.00890.03170.0404
550.07420.19950.1209
560.08280.0336-0.0176
570.00070.03820.0309
58-0.0443-0.0707-0.0719
59-0.0013-0.04220.0273
600.0250-0.0296-0.0885
610.0396-0.00270.0439
62-0.0745-0.0545-0.0858
630.24460.16080.1968
640.07510.00870.0242
65-0.0853-0.1023-0.0748
660.17450.08700.1929
67-0.0608-0.0926-0.0885
680.15160.06340.1421
69-0.0793-0.0896-0.0840
700.03040.04350.0295
71-0.0100-0.0448-0.0300
72-0.0994-0.1104-0.0888
730.03740.0263-0.0497
74-0.1312-0.1267-0.1003
75-0.01620.03280.1302
760.08600.19440.1569
77-0.0239-0.0410-0.0184
780.0395-0.0550-0.0817
79-0.0659-0.0463-0.0796
800.0153-0.0697-0.0301
81-0.0229-0.0738-0.0831
82-0.0050-0.0753-0.0878
83-0.0396-0.0539-0.0714
840.0131-0.02570.0412
85-0.0503-0.0486-0.0845
860.0968-0.04050.1168
87-0.0930-0.0108-0.0308
880.0860-0.0307-0.0026
890.19740.12210.1423
90-0.0471-0.1135-0.0995
910.26490.27480.2964
92-0.1140-0.1293-0.0891
930.02340.08960.0498
940.0049-0.0519-0.0789
95-0.0057-0.0896-0.0842
96-0.02580.04220.0527
97-0.0215-0.0649-0.0121
98-0.09430.07890.1000
99-0.0627-0.0835-0.0966
1000.11270.02720.0064
101-0.0688-0.0057-0.0367
1020.0656-0.0501-0.0300
103-0.1242-0.0997-0.0969
104-0.0981-0.0962-0.0897
105-0.0974-0.0967-0.0885
106-0.1038-0.0268-0.0811
1070.02440.13640.0847
1080.08280.04480.0487
109-0.0529-0.0860-0.0900
1100.02020.01960.0619
111-0.1293-0.1191-0.1064
112-0.1000-0.0870-0.0905
1130.03980.22090.1431
114-0.1267-0.1135-0.0988
115-0.0548-0.0583-0.0319
116-0.0293-0.0667-0.0058
117-0.0774-0.1394-0.1068
118-0.0406-0.0702-0.0894
1190.11970.18930.1194
1200.0653-0.01060.0008
121-0.0834-0.1033-0.0930
122-0.0122-0.0865-0.0238
123-0.0834-0.0077-0.0164
1240.0904-0.0154-0.0583
1250.04460.06390.0401
126-0.0548-0.0896-0.0791
1270.0624-0.02520.0548
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.0404-0.0338-0.0547
10.0040-0.0267-0.0307
2-0.0269-0.0595-0.0116
3-0.1127-0.0571-0.0790
40.06330.10300.0482
5-0.1465-0.1297-0.0519
60.01430.0710-0.0638
7-0.0157-0.1516-0.0503
8-0.02900.0335-0.0082
90.13700.12970.0997
10-0.0392-0.0769-0.0521
11-0.0392-0.01490.0520
12-0.00030.0651-0.0375
13-0.0685-0.0528-0.0439
14-0.05920.0230-0.0015
150.03640.12010.0901
160.09040.07370.0219
170.05110.04700.1395
18-0.01870.0204-0.0505
190.01530.06490.1340
200.04610.13130.0535
210.0108-0.0131-0.0196
22-0.0438-0.0510-0.0022
23-0.0452-0.0422-0.0370
240.16650.14840.1580
25-0.0549-0.0368-0.0913
26-0.0628-0.0201-0.0500
27-0.0459-0.0111-0.0385
280.0751-0.00310.0134
29-0.00730.01850.0774
300.15790.10780.1146
310.07850.07740.1211
32-0.0637-0.0398-0.0511
33-0.0312-0.0790-0.0446
340.10080.06430.0407
350.02900.09610.0919
360.04350.0254-0.0525
37-0.0590-0.0726-0.0343
38-0.2188-0.1687-0.0689
39-0.0423-0.0395-0.0326
40-0.0378-0.04140.0421
410.0077-0.02960.0492
420.01250.0139-0.0076
43-0.1589-0.0454-0.0587
44-0.0713-0.0606-0.0598
45-0.0523-0.00260.0549
460.05020.01910.0622
47-0.00020.09870.0253
48-0.0253-0.0366-0.0530
490.05900.01550.0845
50-0.0106-0.0027-0.0656
51-0.0096-0.0379-0.0197
520.0459-0.00340.0091
53-0.02000.02140.0463
540.00070.03070.0589
550.06710.05950.0383
56-0.05400.03120.0485
570.01640.0360-0.0350
58-0.0609-0.0699-0.0477
590.08180.04880.0980
60-0.0499-0.0694-0.0520
61-0.0369-0.0338-0.0461
62-0.0206-0.0398-0.0436
63-0.0837-0.0358-0.0540
640.05970.01470.1212
650.08850.09340.0494
66-0.02100.0018-0.0222
670.02520.0939-0.0241
680.02880.0090-0.0327
690.0047-0.0184-0.0489
70-0.0080-0.0408-0.0720
710.11410.14200.2117
72-0.0169-0.0148-0.0490
730.09130.03120.0506
74-0.0321-0.0291-0.0536
75-0.01260.00590.0117
760.0011-0.0769-0.0473
77-0.1446-0.1329-0.0542
780.07180.07210.1827
790.08610.06620.1061
80-0.0732-0.0774-0.0528
810.0045-0.0422-0.0541
82-0.0875-0.0622-0.0477
83-0.0913-0.0544-0.0441
840.0206-0.03240.0540
85-0.0932-0.0715-0.0530
86-0.00900.02280.0479
87-0.04780.0057-0.0331
88-0.0326-0.0747-0.0424
890.04760.05180.0967
90-0.1322-0.0150-0.0526
910.07560.07370.0156
92-0.0291-0.0114-0.0458
93-0.0526-0.0998-0.0319
94-0.0614-0.0574-0.0346
95-0.0728-0.0526-0.0553
96-0.0259-0.0552-0.0474
97-0.0108-0.0318-0.0625
980.07510.09980.0140
99-0.0478-0.0267-0.0663
100-0.0086-0.02260.0954
1010.0010-0.00440.0378
1020.0033-0.0239-0.0592
103-0.01330.0335-0.0157
104-0.0932-0.0891-0.0495
105-0.0208-0.01510.0066
106-0.1165-0.0624-0.0409
107-0.0247-0.0076-0.0475
1080.02190.1094-0.0394
109-0.0818-0.0614-0.0510
110-0.0476-0.0403-0.0387
1110.0011-0.00600.0764
112-0.0718-0.0600-0.0428
113-0.0889-0.0587-0.0364
114-0.0761-0.0811-0.0580
115-0.0583-0.0443-0.0376
116-0.02590.0222-0.0505
1170.05640.07420.0784
118-0.0690-0.1025-0.0505
1190.0402-0.01170.0013
120-0.0112-0.0560-0.0676
1210.06020.15270.1307
1220.05610.05020.0352
123-0.0009-0.0649-0.0047
124-0.0210-0.0100-0.0529
1250.0248-0.03390.0475
1260.12650.03790.0934
1270.0030-0.0080-0.0518
128-0.0258-0.0088-0.0400
129-0.0366-0.01570.0115
130-0.0861-0.0993-0.0533
1310.02710.01210.0394
1320.0326-0.0390-0.0562
1330.0125-0.04720.1036
1340.12130.11740.0378
1350.0182-0.0472-0.0313
136-0.0723-0.0758-0.0568
1370.0116-0.00080.0356
1380.05800.09450.0955
139-0.03160.0998-0.0290
1400.0419-0.02170.0479
141-0.0675-0.0769-0.0525
142-0.0718-0.0298-0.0458
1430.01880.06650.0480
144-0.0583-0.0438-0.0553
145-0.0447-0.02620.0342
1460.06610.08700.1260
147-0.0338-0.0263-0.0471
1480.0006-0.0073-0.0545
149-0.0870-0.01610.0086
1500.04070.04560.0109
151-0.06940.0041-0.0137
152-0.0190-0.0231-0.0588
1530.0468-0.00580.0486
154-0.1004-0.0568-0.0517
155-0.0590-0.0173-0.0692
1560.0483-0.0002-0.0290
1570.11080.05150.0614
158-0.0385-0.0822-0.0376
1590.02770.0274-0.0007
160-0.0647-0.0758-0.0504
161-0.0259-0.0822-0.1160
162-0.0559-0.0054-0.0615
163-0.00430.00260.0392
164-0.0380-0.0520-0.0466
1650.0205-0.07470.0490
166-0.0604-0.0097-0.0617
1670.11700.07310.1370
1680.06710.11370.0349
1690.11030.18360.0868
170-0.0248-0.03520.0044
171-0.0766-0.0574-0.0467
1720.06940.03380.0465
173-0.0302-0.0558-0.0621
1740.01690.0811-0.0489
175-0.0918-0.1030-0.0878
176-0.00730.05500.1155
177-0.0323-0.0360-0.0437
178-0.0642-0.0502-0.0471
179-0.0271-0.0127-0.0653
1800.09800.14090.1021
181-0.0273-0.00930.0621
182-0.0476-0.0649-0.0458
183-0.0380-0.00370.0066
184-0.0340-0.0167-0.0538
1850.13600.15160.0888
186-0.0480-0.0392-0.0639
1870.08560.09550.0757
1880.0285-0.01070.0346
1890.15120.13980.1604
190-0.0195-0.0294-0.0178
1910.0079-0.02160.0273
192-0.0454-0.0528-0.0641
193-0.0291-0.0091-0.0558
194-0.00870.0467-0.0331
195-0.0423-0.0180-0.0504
1960.06850.0400-0.0453
1970.06750.0024-0.0165
198-0.0161-0.0347-0.0512
199-0.00180.09130.0236
200-0.0302-0.0606-0.0613
2010.0143-0.00140.0349
202-0.1089-0.1318-0.1103
203-0.0086-0.0276-0.0661
2040.11370.10940.1132
2050.0889-0.00830.1042
206-0.0350-0.0288-0.0305
207-0.0044-0.03900.0185
208-0.0870-0.0259-0.0550
2090.0585-0.02520.0242
210-0.00070.0319-0.0439
211-0.0526-0.0539-0.0407
2120.06520.06140.1337
2130.0127-0.05520.0232
2140.00840.02140.0852
2150.05400.05740.0204
216-0.0799-0.0382-0.0634
217-0.02000.00320.0270
218-0.0942-0.1537-0.0615
219-0.0614-0.0432-0.0503
220-0.0554-0.0438-0.0526
221-0.04920.0030-0.0460
222-0.0521-0.0627-0.0621
223-0.0604-0.0276-0.0210
224-0.0407-0.0406-0.0233
225-0.0471-0.0303-0.0441
2260.04990.12490.0876
2270.11320.16870.0718
2280.05350.0292-0.0504
2290.0704-0.0159-0.0581
230-0.0528-0.0555-0.0439
2310.10420.01110.1741
2320.0307-0.0020-0.0503
233-0.1313-0.0870-0.0694
2340.07560.0085-0.0450
2350.07180.02100.0513
236-0.0195-0.0475-0.0455
237-0.0380-0.0390-0.0541
238-0.0438-0.0322-0.0312
239-0.0190-0.0622-0.0548
2400.0545-0.0199-0.0540
2410.0354-0.00650.0087
242-0.0894-0.0472-0.0781
243-0.0842-0.0833-0.0681
244-0.00290.00100.0320
245-0.0138-0.04750.0706
2460.0537-0.0606-0.0429
247-0.0661-0.0406-0.0446
248-0.0685-0.0681-0.0606
249-0.0099-0.0096-0.0406
Source: M2_RECENT_DPO_LORA_SPEAKER_EMOTION_STUDY_2026-10-02.html · SHA-256 91445a82934d6abe…
g04GermanDPO acting study

HVSAC-030 · de · targets: Pride, Confusion, Amusement

Generated TTS prediction · S3 base · C01 · 16.4 s

Die Suppe ist ausgezeichnet, obwohl der Löffel eindeutig eine fragwürdige Haltung hat. Dieser Tisch steht außerdem schief, oder vielleicht ist es die Kunst an der Wand. Bitte bringen Sie mir noch einen Kaffee, ich muss weiter urteilen.

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Teasing
z +3.03 / +3.37raw 1.65 / 1.84
Fear
z +1.92 / +1.90raw 1.19 / 1.17
Contentment
z +1.64 / +1.83raw 0.84 / 0.94
Relief
z +1.74 / +1.49raw 1.22 / 1.05
Fatigue Exhaustion
z +1.35 / +1.53raw 0.75 / 0.85
Amusement
z +1.49 / +1.39raw 1.32 / 1.23
Intoxication Altered States of Consciousness
z +1.31 / +0.67raw 0.71 / 0.35

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Narrator style (S_NARR)
z +1.33 / +1.14raw 3.22 / 2.96
Content appropriateness (EXPL)
z +1.28 / +1.18raw 0.50 / 0.47
Volatility (VOLT)
z +1.17 / +1.10raw 3.62 / 3.56
Cartoonish style (S_CART)
z +1.15 / +0.99raw 3.17 / 2.95
Teacher or didactic style (S_TECH)
z +0.98 / +1.10raw 1.85 / 1.99
Vulnerability (VULN)
z +0.67 / +0.91raw 3.31 / 3.54
Perceived gender (GEND)
z +0.84 / +0.89raw 5.39 / 5.46

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z -0.22 / -0.15raw 1.98 / 2.08
Vocal-burst blend (0–10)z -1.71 / -1.57raw 1.99 / 2.38
Content enjoyment (CE)z +0.80 / +0.80raw 5.61 / 5.61
Production quality (PQ)z +0.81 / +0.63raw 7.75 / 7.60
Speech signal quality, calibratedz -0.35 / -0.20raw 3.32 / 3.39
Background quality, calibratedz +0.08 / +0.17raw 3.46 / 3.55
Overall quality, calibratedz +0.18 / +0.17raw 2.95 / 2.94
P.808 MOS estimatez +0.50 / +0.17raw 3.67 / 3.50
score content enjoymentz +1.10 / +0.97raw 4.84 / 4.80
score overall qualityz +0.38 / +0.25raw 2.84 / 2.78
Background Noisez -0.35 / +0.02raw 0.71 / 0.81
Tempo (TEMP)z -1.13 / -0.82raw 0.77 / 1.14
Arousal (AROU)z -1.21 / -0.95raw 0.98 / 1.27
Valence (VALN)z -0.61 / -0.49raw 1.65 / 1.78
Recording quality (RCQL)z +0.09 / -0.15raw 2.84 / 2.62
Background noise (BKGN)z +0.32 / +0.34raw 2.94 / 2.96
talking speedz -0.58 / -0.08raw 6.86 / 8.73
Characters per second (CPS): 14.29 raw · z +1.40

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.7870.839
Identity · 250D0.7640.843

Vocal-burst regions

Tiny
Base
0 s16.4 s

Verified burst intervals are unavailable for this clip.

Tiny: 1 regions, count-head estimate 1.02; Base: 1 regions, count-head estimate 0.91.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
Tiny15.145.480.34
Base15.185.480.30
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.127+0.210.129+0.210.0070.570
1Amusementemo_Amusement1.322+1.491.227+1.39-0.0010.885
2Angeremo_Anger-0.077-0.13-0.022-0.040.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.107+0.18-0.009-0.030.0050.572
4Aweemo_Awe0.002+0.00-0.021-0.050.0000.427
5Bitternessemo_Bitterness-0.022-0.080.014+0.000.0130.415
6Concentrationemo_Concentration0.808+0.230.953+0.430.6340.741
7Confusionemo_Confusion0.457+0.850.297+0.550.0010.540
8Contemplationemo_Contemplation0.461+0.630.672+0.990.0820.598
9Contemptemo_Contempt0.021+0.050.062+0.14-0.0000.464
10Contentmentemo_Contentment0.840+1.640.941+1.830.0000.514
11Disappointmentemo_Disappointment-0.088-0.130.005+0.010.0000.700
12Disgustemo_Disgust-0.008-0.030.019+0.020.0070.489
13Distressemo_Distress0.307+0.370.314+0.380.0000.821
14Doubtemo_Doubt0.514+1.090.450+0.950.0060.466
15Elationemo_Elation0.227+0.300.355+0.470.0120.728
16Embarrassmentemo_Embarrassment0.069+0.060.028-0.040.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.492+0.530.667+0.720.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.754+1.350.853+1.530.0280.539
19Fearemo_Fear1.187+1.921.172+1.900.0150.610
20Helplessnessemo_Helplessness0.307+0.450.163+0.24-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.475+0.471.019+1.190.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.140-0.010.202+0.070.1490.705
23Infatuationemo_Infatuation0.016+0.060.018+0.06-0.0150.533
24Interestemo_Interest1.740+0.171.945+0.391.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness0.709+1.310.347+0.67-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.056-0.010.053-0.010.0590.442
27Longingemo_Longing0.380+0.460.568+0.760.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.154+0.180.088+0.060.0550.534
29Painemo_Pain0.059+0.11-0.003-0.00-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.707+1.120.772+1.230.0170.614
31Prideemo_Pride0.005+0.01-0.001-0.010.0020.509
32Reliefemo_Relief1.221+1.741.046+1.490.0330.681
33Sadnessemo_Sadness0.256+0.340.351+0.460.0000.759
34Sexual Lustemo_Sexual_Lust0.292+0.530.281+0.510.0300.494
35Shameemo_Shame0.013-0.000.022+0.020.0130.510
36Sournessemo_Sourness0.031+0.080.007+0.03-0.0050.439
37Teasingemo_Teasing1.653+3.031.838+3.370.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.151+0.160.178+0.200.0410.697
39Triumphemo_Triumph0.045+0.10-0.010-0.01-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg2.885-0.082.526-0.372.9841.238
41Arousal (AROU)vn_AROU_reg0.976-1.211.266-0.952.3591.147
42Arousal shift (ARSH)vn_ARSH_reg2.932+0.412.911+0.382.5800.867
43Vocal attack (ATCK)vn_ATCK_reg0.771-1.071.044-0.862.1621.295
44Background noise (BKGN)vn_BKGN_reg2.944+0.322.962+0.342.7030.761
45Brightness (BRGT)vn_BRGT_reg1.277-0.881.441-0.681.9910.813
46Speech chunking (CHNK)vn_CHNK_reg1.001-0.681.181-0.541.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg1.267-0.791.600-0.442.0180.945
48Cognitive load (COGL)vn_COGL_reg2.298+0.192.396+0.292.0941.061
49Dynamic arc (DARC)vn_DARC_reg1.480-0.101.773+0.111.6151.380
50Disfluency (DFLU)vn_DFLU_reg3.147+0.563.295+0.682.4881.184
51Emphasis (EMPH)vn_EMPH_reg2.813+0.132.905+0.202.6071.524
52Esthetics (ESTH)vn_ESTH_reg2.138-0.342.135-0.342.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.500+1.280.467+1.180.0860.322
54Focus (FOCS)vn_FOCS_reg1.993-0.321.613-0.662.3441.102
55Fullness (FULL)vn_FULL_reg1.575-0.231.684-0.111.7780.889
56Perceived gender (GEND)vn_GEND_reg5.387+0.845.461+0.894.0551.588
57Harmonicity (HARM)vn_HARM_reg1.443-1.241.738-0.892.4800.836
58Metallic character (METL)vn_METL_reg0.681-0.980.880-0.771.6210.958
59Pitch range (RANG)vn_RANG_reg2.935+0.023.226+0.352.9140.897
60Recording quality (RCQL)vn_RCQL_reg2.836+0.092.621-0.152.7540.888
61Vocal register (REGS)vn_REGS_reg0.466-0.580.780-0.331.1981.268
62Respiration (RESP)vn_RESP_reg3.390+0.673.298+0.592.6801.056
63Roughness (ROUG)vn_ROUG_reg2.650+0.702.250+0.242.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg2.582+0.532.448+0.372.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg1.616-0.651.876-0.332.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg1.367-0.541.640-0.271.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg1.349-0.101.257-0.211.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg1.666-1.072.177-0.592.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg1.574-0.821.758-0.572.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg0.973-1.371.093-1.222.0700.803
71Smoothness (SMTH)vn_SMTH_reg1.709-0.401.299-0.772.1431.093
72Stance (STNC)vn_STNC_reg2.116-0.452.194-0.392.6331.137
73Speech structure (STRU)vn_STRU_reg1.419-0.751.478-0.692.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg2.438+0.762.082+0.501.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg1.006-0.081.252+0.161.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg3.166+1.152.952+0.991.5601.401
77Casual style (S_CASU)vn_S_CASU_reg2.988-0.633.197-0.473.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg2.835+0.183.046+0.342.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg3.220+0.433.284+0.492.7211.148
80Formal style (S_FORM)vn_S_FORM_reg1.009-0.181.098-0.101.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg3.491+0.523.357+0.412.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg3.224+1.332.955+1.141.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.429+0.040.429+0.040.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg2.761+0.312.954+0.462.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg1.520+0.241.362+0.121.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg3.305+0.603.039+0.412.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg1.851+0.981.994+1.100.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg2.697+0.682.363+0.471.5971.618
89Tempo (TEMP)vn_TEMP_reg0.774-1.131.139-0.822.1171.190
90Tension (TENS)vn_TENS_reg2.066+0.602.102+0.641.5590.852
91Valence (VALN)vn_VALN_reg1.645-0.611.784-0.492.3281.118
92Valence shift (VALS)vn_VALS_reg2.050-0.432.071-0.412.5181.079
93Velocity flux (VFLX)vn_VFLX_reg-0.084-1.940.161-1.732.2561.208
94Volatility (VOLT)vn_VOLT_reg3.621+1.173.560+1.102.5860.886
95Vulnerability (VULN)vn_VULN_reg3.313+0.673.545+0.912.6480.984
96Warmth (WARM)vn_WARM_reg3.242+0.732.998+0.462.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_61.978-0.222.085-0.152.3321.621
98Vocal-burst blend (0–10)blend_0_101.994-1.712.376-1.576.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.549-1.620.601-1.461.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age2.976-0.942.545-2.413.2500.293
101Arousaleiv_extra:Arousal1.413+0.291.567+0.641.2890.435
102Authenticityeiv_extra:Authenticity2.717+0.322.746+0.522.6670.153
103Background Noiseeiv_extra:Background_Noise0.714-0.350.813+0.020.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.118+0.261.220+0.591.0400.303
105Gendereiv_extra:Gender-0.372+0.25-0.207+0.41-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.717-0.441.751-0.311.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive1.886+0.062.043+0.391.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.195+0.152.215+0.212.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous1.083+0.451.142+0.620.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.492+0.660.698+1.300.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.277-0.300.316-0.180.3760.324
112Valenceeiv_extra:Valence0.148+0.380.477+1.240.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.482-0.531.493-0.511.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold0.400+1.160.484+1.440.0510.300
115score background qualityeiv_extra:score_background_quality3.396+0.143.396+0.143.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.840+1.104.802+0.974.5070.303
117score overall qualityeiv_extra:score_overall_quality2.842+0.382.777+0.252.6520.507
118score speech qualityeiv_extra:score_speech_quality1.721-0.591.718-0.621.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.614+0.805.614+0.804.7991.018
120Content usefulness (CU)audiobox:CU6.514+0.866.713+1.065.6491.001
121Production complexity (PC)audiobox:PC2.753+0.262.824+0.302.3311.617
122Production quality (PQ)audiobox:PQ7.754+0.817.596+0.637.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw3.750-0.223.781-0.183.9000.669
124Background quality, rawdnsmos:BAK_raw3.472+0.233.643+0.383.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.297+0.263.302+0.273.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.315-0.353.388-0.203.4880.501
127Background quality, calibrateddnsmos:BAK3.462+0.083.546+0.173.3890.928
128Overall quality, calibrateddnsmos:OVRL2.946+0.182.943+0.172.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.671+0.503.505+0.173.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.705+1.090.646+1.000.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.779-0.300.643-0.550.9470.554
132Agevoiceclap_attribute:Age2.491-1.352.518-1.303.2290.547
133Amusementvoiceclap_attribute:Amusement1.276+1.550.877+0.850.3920.572
134Angervoiceclap_attribute:Anger0.868-0.250.961-0.060.9920.506
135Arousalvoiceclap_attribute:Arousal1.638+0.351.644+0.361.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise1.216+1.481.122+1.280.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.699-0.432.832+0.372.7700.167
138Awevoiceclap_attribute:Awe0.353-0.720.416-0.470.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.690+0.470.786+0.830.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.492-0.420.602+0.040.5920.236
141Concentrationvoiceclap_attribute:Concentration0.690-0.990.827-0.731.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant0.914-0.330.764-0.651.0690.468
143Confusionvoiceclap_attribute:Confusion0.963+0.960.829+0.560.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.563-0.540.366-1.020.7800.406
145Contemptvoiceclap_attribute:Contempt0.778+0.210.574-0.400.7090.339
146Contentmentvoiceclap_attribute:Contentment0.437-0.490.639+0.170.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.690-0.240.762+0.010.7600.291
148Disgustvoiceclap_attribute:Disgust0.765+0.680.794+0.820.6250.207
149Distressvoiceclap_attribute:Distress1.000+0.211.137+0.600.9270.350
150Doubtvoiceclap_attribute:Doubt0.484-0.270.465-0.370.5350.188
151Elationvoiceclap_attribute:Elation0.573+0.080.587+0.120.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.412+0.340.221-0.270.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.702+0.140.563-0.380.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.471+0.040.471+0.040.4620.261
155Fearvoiceclap_attribute:Fear1.005+1.821.009+1.840.6180.212
156Gendervoiceclap_attribute:Gender-0.009+0.140.213+0.36-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.840+0.680.928+1.010.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.567-0.731.553-0.771.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.412-0.740.605-0.250.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability0.709-0.630.796-0.471.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.724+0.350.572-0.110.6080.332
162Interestvoiceclap_attribute:Interest1.762+0.371.686+0.201.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness1.264+1.640.921+0.900.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.698+0.720.668+0.590.5230.245
165Longingvoiceclap_attribute:Longing0.398-0.640.362-0.800.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice1.172+0.981.012+0.520.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive1.583-0.321.613-0.271.7750.594
168Painvoiceclap_attribute:Pain0.622+1.100.707+1.490.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.652+0.490.556+0.180.5030.301
170Pridevoiceclap_attribute:Pride0.626-0.440.634-0.420.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.203-0.342.213-0.312.3130.328
172Reliefvoiceclap_attribute:Relief1.238+0.521.456+1.041.0250.415
173Sadnessvoiceclap_attribute:Sadness0.660-0.290.779+0.090.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous1.394+1.041.079+0.480.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust1.136+1.200.804+0.400.6350.419
176Shamevoiceclap_attribute:Shame0.373-0.540.577+0.180.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.396+0.660.212+0.150.1600.360
178Sournessvoiceclap_attribute:Sourness0.601-0.360.639-0.230.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.343-0.780.336-0.800.6230.359
180Teasingvoiceclap_attribute:Teasing1.580+2.741.065+1.380.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.241-0.271.323-0.111.3840.529
182Triumphvoiceclap_attribute:Triumph1.068+0.981.129+1.150.7160.359
183Valencevoiceclap_attribute:Valence0.240+0.39-0.024-0.03-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached1.357-0.511.408-0.411.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold0.152+0.550.008+0.19-0.0660.400
186durationvoiceclap_attribute:duration17.128+0.9817.128+0.9810.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.456+0.423.624+0.763.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.907+1.714.920+1.804.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.797+0.392.878+0.592.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.891-0.001.928+0.531.8910.069
191talking speedvoiceclap_attribute:talking_speed6.858-0.588.727-0.089.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.04210.07620.0727
1-0.0208-0.1238-0.0892
20.20320.19030.0704
30.0231-0.01740.0320
40.08510.0493-0.0440
5-0.0446-0.0905-0.1136
60.08420.10700.0800
7-0.0723-0.0674-0.0822
8-0.05080.02560.0238
9-0.0387-0.01300.0228
100.20510.23070.1564
11-0.0677-0.0947-0.1050
120.04230.01510.0289
130.0063-0.0090-0.0141
140.13090.07200.0296
15-0.1167-0.1238-0.0804
16-0.0838-0.1137-0.1111
17-0.0405-0.0564-0.0933
18-0.0865-0.1027-0.0828
190.13370.03200.0324
20-0.0916-0.0985-0.0899
21-0.0090-0.00820.0358
220.21240.19450.2688
230.01130.03640.0744
24-0.0641-0.0115-0.0381
25-0.0737-0.0901-0.0789
26-0.0714-0.0568-0.0842
27-0.0943-0.0594-0.0837
28-0.00810.02490.0421
290.04620.09430.0404
30-0.1611-0.1044-0.0966
310.0391-0.0535-0.0198
320.17300.24090.1849
33-0.1140-0.0893-0.0989
34-0.0851-0.0392-0.0805
35-0.1199-0.0459-0.0501
360.05740.01670.0277
37-0.0760-0.1095-0.1063
380.0401-0.0392-0.0753
390.07370.06440.0214
40-0.0256-0.0314-0.0865
41-0.0259-0.06020.0089
420.05740.04380.0891
43-0.0929-0.0556-0.0923
44-0.1172-0.0821-0.0862
450.05220.0347-0.0009
46-0.03940.0201-0.0142
470.05080.07450.0584
48-0.0536-0.1145-0.0876
49-0.0035-0.04860.0267
50-0.0165-0.0474-0.0199
510.09520.08210.0831
52-0.0451-0.0328-0.1020
53-0.0719-0.0884-0.0916
54-0.1282-0.2089-0.1005
55-0.0513-0.0859-0.0864
560.0046-0.0459-0.0784
57-0.0829-0.1314-0.0984
580.0391-0.0322-0.0147
59-0.0298-0.05030.0411
60-0.0068-0.0023-0.0866
610.05380.07920.1005
620.07000.16000.2194
63-0.1511-0.1752-0.0853
64-0.00080.01790.0015
650.14280.06190.1176
660.15380.12970.0141
670.25820.21900.2749
680.01400.02760.0356
690.16940.09640.1219
70-0.0948-0.0585-0.0930
71-0.0522-0.0733-0.0995
720.22520.21730.2263
730.06360.00720.0071
74-0.0437-0.04510.0326
750.03320.0045-0.0352
76-0.0861-0.0994-0.0871
77-0.1126-0.1086-0.0820
780.03460.0143-0.0202
79-0.0038-0.0095-0.0385
800.12080.05520.1752
810.01990.09900.0790
820.02510.0172-0.0233
830.09980.08970.1032
840.06450.04110.0652
85-0.0269-0.00580.0596
860.0396-0.0318-0.0259
87-0.0700-0.0703-0.0908
880.02290.01010.0427
89-0.0336-0.0278-0.0671
900.07740.02950.0291
91-0.0938-0.0851-0.0825
920.22430.13730.0998
93-0.0980-0.1482-0.0878
94-0.0137-0.02760.0419
95-0.0584-0.1322-0.0964
96-0.1048-0.0888-0.0930
97-0.0742-0.1048-0.0665
98-0.0796-0.0872-0.0890
990.05950.02490.0029
1000.09110.03520.1849
1010.05200.10230.0477
1020.14830.16250.0787
1030.18310.09310.1012
1040.02560.06150.0914
1050.0373-0.00850.0792
1060.03300.00420.0053
1070.0403-0.01490.0002
1080.02560.04360.0921
109-0.0314-0.0779-0.0030
1100.04490.0728-0.0704
1110.06730.0476-0.0072
112-0.0251-0.0260-0.0885
113-0.0481-0.0079-0.0064
1140.01200.15830.0180
115-0.0227-0.0206-0.0851
116-0.0341-0.0678-0.0770
117-0.03360.11120.1434
118-0.0012-0.0585-0.0874
119-0.1131-0.0952-0.0868
1200.05260.07160.1052
121-0.0057-0.02970.0162
122-0.0151-0.0202-0.0462
123-0.0506-0.02570.0481
1240.0221-0.0472-0.0908
125-0.1172-0.0539-0.0895
126-0.0323-0.0227-0.0774
127-0.1566-0.1347-0.0953
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.0266-0.0530-0.0444
1-0.0803-0.0484-0.0390
2-0.01260.0204-0.0293
30.0522-0.0463-0.0314
4-0.1000-0.0766-0.0367
5-0.0613-0.01680.0486
6-0.00170.07910.0067
70.19260.12570.1378
80.07040.06380.0704
90.00510.0193-0.0013
10-0.00120.07480.0497
11-0.0624-0.0494-0.0546
12-0.1043-0.0635-0.0441
130.13940.10850.0728
14-0.0482-0.0846-0.0575
150.25830.19130.1976
160.06530.06500.0530
17-0.02700.0118-0.0484
180.09920.01810.0919
19-0.0916-0.0745-0.0640
200.0018-0.0512-0.0404
21-0.0496-0.0330-0.0374
220.0146-0.0101-0.0113
23-0.0012-0.0466-0.0390
24-0.0405-0.0277-0.0344
25-0.0522-0.01820.0068
26-0.01230.08400.0611
27-0.0281-0.0583-0.0328
28-0.0077-0.0075-0.0436
29-0.0312-0.0570-0.0388
30-0.0401-0.0521-0.0351
310.0452-0.0616-0.0507
320.03340.07140.0203
330.06970.17170.1608
34-0.1065-0.0589-0.0482
350.0001-0.0320-0.0519
36-0.0327-0.0277-0.0426
37-0.0093-0.0018-0.0279
38-0.0660-0.0684-0.0559
39-0.0201-0.0671-0.0193
40-0.0544-0.0423-0.0467
41-0.00230.0162-0.0463
420.03230.04050.0412
430.03520.03420.0252
44-0.00230.0322-0.0133
450.0085-0.0319-0.0463
46-0.0500-0.0576-0.0592
47-0.0379-0.0302-0.0487
48-0.0241-0.0126-0.0430
49-0.07190.02410.0178
500.05440.06710.0775
51-0.03030.07600.0250
52-0.0558-0.0702-0.0398
530.05980.06650.0221
540.00240.02050.0217
550.08430.07080.0488
560.0164-0.01000.0681
57-0.03650.0057-0.0018
580.0330-0.0111-0.0387
590.09850.04970.0618
600.10220.10420.1578
610.06420.05580.0305
62-0.03980.0092-0.0354
630.00970.01530.0528
640.00160.0221-0.0449
65-0.0985-0.0714-0.0492
660.07040.06220.0383
67-0.00810.0379-0.0017
680.02830.02240.0194
690.08720.06100.0734
70-0.0452-0.0742-0.0585
71-0.0409-0.0822-0.0422
72-0.0500-0.0675-0.0447
73-0.0367-0.0288-0.0720
74-0.0432-0.0435-0.0435
75-0.0254-0.01800.0017
76-0.0396-0.0733-0.0384
770.0077-0.0549-0.0221
78-0.00710.02420.0311
790.09490.04780.0365
80-0.0471-0.0668-0.0428
810.09780.10550.0378
820.13350.10490.1763
83-0.0345-0.0320-0.0358
840.06750.02970.0407
85-0.0396-0.0441-0.0066
86-0.0821-0.0561-0.0385
870.0098-0.0274-0.0344
880.09010.10790.0659
890.05950.06710.0216
900.03170.08770.0425
910.0056-0.0199-0.0356
92-0.0432-0.0696-0.0372
93-0.04600.0091-0.0489
940.0001-0.00840.0222
950.18240.17050.0784
960.04850.05760.0273
97-0.0337-0.0227-0.0176
980.12330.09810.0295
99-0.0139-0.0475-0.0539
1000.07700.09500.1249
1010.06130.0506-0.0228
1020.0058-0.00320.0152
103-0.0142-0.0244-0.0712
104-0.0002-0.0366-0.0437
105-0.0722-0.0763-0.0560
106-0.0503-0.0503-0.0332
1070.07190.0262-0.0131
108-0.0518-0.0273-0.0320
1090.08830.16800.1503
110-0.0573-0.0414-0.0513
1110.0058-0.0035-0.0200
112-0.0272-0.0238-0.0348
113-0.0766-0.0742-0.0648
1140.04360.05950.1107
1150.0401-0.0035-0.0305
1160.01780.0248-0.0412
117-0.0144-0.0058-0.0339
118-0.1167-0.1049-0.0562
1190.05550.13370.1693
120-0.01090.0247-0.0376
121-0.0171-0.0466-0.0367
1220.0135-0.0027-0.0406
1230.09630.11280.1555
124-0.0156-0.0451-0.0429
125-0.1306-0.0675-0.0418
126-0.0392-0.0231-0.0429
127-0.0562-0.0834-0.0443
128-0.01990.01030.0283
129-0.0317-0.0487-0.0457
130-0.02080.0205-0.0432
131-0.0500-0.0385-0.0377
132-0.0281-0.0284-0.0457
133-0.02960.01770.0328
134-0.02660.02390.0494
135-0.00290.06470.0347
136-0.00150.0126-0.0461
137-0.0522-0.0213-0.0400
138-0.0744-0.0675-0.0380
1390.04960.0176-0.0164
1400.01290.01320.1018
141-0.0212-0.0227-0.0426
142-0.0390-0.0328-0.0372
143-0.1036-0.0586-0.0587
144-0.0204-0.0327-0.0449
145-0.0682-0.0524-0.0547
146-0.05760.0299-0.0022
147-0.0100-0.0146-0.0016
1480.02130.0328-0.0474
1490.08280.14040.1710
1500.0188-0.0071-0.0424
1510.08060.06190.0782
1520.04120.0093-0.0423
1530.05650.07570.0718
1540.05950.08030.0978
1550.04400.0247-0.0533
1560.0135-0.0136-0.0431
1570.00340.0589-0.0137
158-0.0440-0.0108-0.0388
159-0.0660-0.0858-0.0504
1600.05580.12750.2253
161-0.0941-0.0322-0.0277
162-0.0055-0.0368-0.0500
163-0.0664-0.0714-0.0479
1640.03410.0026-0.0378
165-0.0387-0.0198-0.0277
1660.03480.0926-0.0107
1670.0843-0.0182-0.0288
168-0.0522-0.0629-0.0609
169-0.0887-0.0021-0.0491
170-0.0279-0.0262-0.0403
1710.01040.0399-0.0049
172-0.0642-0.0215-0.0499
1730.02630.06410.0217
174-0.0482-0.0926-0.0442
175-0.0956-0.1036-0.0713
176-0.0624-0.0644-0.0366
177-0.0376-0.0333-0.0355
1780.15690.17290.1722
1790.10430.10360.0551
180-0.03430.0014-0.0022
181-0.02230.00130.0347
1820.03030.04780.0084
183-0.0098-0.0478-0.0338
184-0.0182-0.0213-0.0437
185-0.0540-0.0299-0.0515
186-0.0653-0.0552-0.0519
187-0.01130.0033-0.0144
1880.00210.06160.0786
1890.00590.04940.0336
1900.06380.09500.0682
191-0.0190-0.0330-0.0132
192-0.0503-0.0445-0.0521
1930.15470.11100.1465
194-0.0257-0.0451-0.0558
195-0.0248-0.0009-0.0191
1960.04310.0293-0.0279
197-0.07810.0012-0.0432
198-0.0272-0.0130-0.0358
199-0.0573-0.03630.0047
200-0.0770-0.0647-0.0497
2010.07370.07330.0622
202-0.0078-0.0082-0.0891
2030.0010-0.00510.0096
2040.07920.11410.1306
2050.10870.09870.1039
2060.04050.0463-0.0118
2070.0693-0.00090.0309
208-0.02590.00370.0031
209-0.01720.02160.0027
2100.13130.22930.1411
211-0.0131-0.0134-0.0331
2120.09340.05270.0502
2130.11600.11220.0950
2140.13210.13670.1581
215-0.01690.0041-0.0362
2160.09700.16800.1172
2170.03500.0089-0.0232
2180.10580.06070.1316
219-0.0872-0.0684-0.0560
220-0.0693-0.0653-0.0475
221-0.0533-0.0173-0.0373
222-0.01510.00070.0559
223-0.04430.0104-0.0044
2240.05470.02640.0224
2250.02020.06500.0152
2260.13210.08280.1364
227-0.0598-0.0690-0.0569
228-0.1116-0.0757-0.0696
2290.02000.07330.0019
230-0.0096-0.0001-0.0438
231-0.0529-0.0311-0.0595
232-0.0235-0.0742-0.0591
233-0.0237-0.0256-0.0613
234-0.0288-0.0402-0.0365
2350.10220.10550.1379
236-0.01420.00630.0052
237-0.0122-0.0095-0.0439
2380.00550.02700.0217
239-0.0361-0.0299-0.0445
240-0.0434-0.0540-0.0439
2410.04090.07970.1390
242-0.0540-0.0396-0.0634
243-0.0544-0.0319-0.0553
2440.0339-0.0043-0.0457
2450.15320.10360.1313
246-0.0576-0.0840-0.0349
247-0.0100-0.0189-0.0362
248-0.0462-0.0647-0.0492
2490.10220.02210.0276
Source: M2_RECENT_DPO_LORA_SPEAKER_EMOTION_STUDY_2026-10-02.html · SHA-256 b4f234db7c315319…
g05EnglishMediathek voice study

mediathek_0047 · en · 1/50

Generated TTS prediction · Rank 2 qkvo · 7.6 s

We really need to make these legal ideas about getting along more solid. It is crucial to get concrete about these principles of coexistence.

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Concentration
z +1.67 / +1.90raw 1.87 / 2.05
Contemplation
z +1.82 / +1.43raw 1.17 / 0.94
Contempt
z +1.63 / +1.22raw 0.76 / 0.56
Sourness
z +1.47 / +1.32raw 0.64 / 0.57
Impatience and Irritability
z +1.39 / +1.33raw 1.13 / 1.09
Disgust
z +1.27 / +0.55raw 0.63 / 0.28
Anger
z +1.23 / +1.10raw 0.75 / 0.68

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Newsreader style (S_NEWS)
z +2.46 / +3.05raw 1.80 / 2.14
Authoritative style (S_AUTH)
z +1.73 / +1.77raw 2.89 / 2.93
Content appropriateness (EXPL)
z +1.29 / +1.37raw 0.50 / 0.53
Speech chunking (CHNK)
z +1.14 / +1.11raw 3.26 / 3.23
Tempo (TEMP)
z +1.10 / +1.05raw 3.43 / 3.37
Metallic character (METL)
z +0.98 / +0.86raw 2.56 / 2.45
Speech structure (STRU)
z +0.78 / +0.96raw 2.88 / 3.06

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z +0.04 / -0.30raw 2.40 / 1.85
Vocal-burst blend (0–10)z -0.87 / -1.03raw 4.29 / 3.86
Content enjoyment (CE)z +0.71 / +0.96raw 5.53 / 5.78
Production quality (PQ)z +0.28 / +0.08raw 7.29 / 7.12
Speech signal quality, calibratedz +0.19 / +0.15raw 3.59 / 3.57
Background quality, calibratedz +0.57 / +0.65raw 3.92 / 3.99
Overall quality, calibratedz +0.61 / +0.63raw 3.21 / 3.22
P.808 MOS estimatez +0.84 / +0.76raw 3.84 / 3.80
score content enjoymentz +0.62 / +1.00raw 4.69 / 4.81
score overall qualityz +0.55 / +0.63raw 2.93 / 2.97
Background Noisez -1.34 / -2.03raw 0.45 / 0.27
Tempo (TEMP)z +1.10 / +1.05raw 3.43 / 3.37
Arousal (AROU)z +0.36 / +0.29raw 2.77 / 2.69
Valence (VALN)z +0.33 / +0.09raw 2.70 / 2.43
Recording quality (RCQL)z -0.53 / -0.43raw 2.28 / 2.37
Background noise (BKGN)z +0.04 / -0.01raw 2.74 / 2.70
talking speedz +3.18 / +3.38raw 20.97 / 21.74
Characters per second (CPS): 18.47 raw · z +2.52

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8830.897
Identity · 250D0.6830.707

Vocal-burst regions

Tiny
Base
0 s7.6 s

Verified burst intervals are unavailable for this clip.

Tiny: 0 regions, count-head estimate 0.01; Base: 0 regions, count-head estimate 0.00.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.032+0.040.049+0.070.0070.570
1Amusementemo_Amusement-0.014-0.01-0.034-0.04-0.0010.885
2Angeremo_Anger0.755+1.230.680+1.100.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.004-0.000.052+0.080.0050.572
4Aweemo_Awe0.043+0.100.037+0.090.0000.427
5Bitternessemo_Bitterness0.288+0.660.272+0.620.0130.415
6Concentrationemo_Concentration1.875+1.672.046+1.900.6340.741
7Confusionemo_Confusion-0.016-0.03-0.005-0.010.0010.540
8Contemplationemo_Contemplation1.173+1.820.938+1.430.0820.598
9Contemptemo_Contempt0.755+1.630.565+1.22-0.0000.464
10Contentmentemo_Contentment0.020+0.040.096+0.190.0000.514
11Disappointmentemo_Disappointment0.246+0.350.131+0.190.0000.700
12Disgustemo_Disgust0.628+1.270.277+0.550.0070.489
13Distressemo_Distress0.117+0.14-0.014-0.020.0000.821
14Doubtemo_Doubt0.188+0.390.048+0.090.0060.466
15Elationemo_Elation0.125+0.150.173+0.220.0120.728
16Embarrassmentemo_Embarrassment-0.010-0.12-0.036-0.180.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.327+0.350.224+0.240.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.158+0.240.117+0.160.0280.539
19Fearemo_Fear0.145+0.21-0.032-0.080.0150.610
20Helplessnessemo_Helplessness0.084+0.13-0.006-0.00-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.773+0.860.883+1.010.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability1.126+1.391.089+1.330.1490.705
23Infatuationemo_Infatuation-0.027-0.02-0.015-0.00-0.0150.533
24Interestemo_Interest2.174+0.652.220+0.701.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness-0.053-0.05-0.064-0.07-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.071+0.030.095+0.080.0590.442
27Longingemo_Longing0.168+0.130.143+0.090.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.361+0.570.190+0.250.0550.534
29Painemo_Pain0.013+0.03-0.019-0.03-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.016-0.00-0.032-0.080.0170.614
31Prideemo_Pride0.134+0.260.376+0.740.0020.509
32Reliefemo_Relief0.052+0.030.031-0.000.0330.681
33Sadnessemo_Sadness0.023+0.03-0.067-0.090.0000.759
34Sexual Lustemo_Sexual_Lust0.090+0.120.019-0.020.0300.494
35Shameemo_Shame0.034+0.040.018+0.010.0130.510
36Sournessemo_Sourness0.641+1.470.575+1.32-0.0050.439
37Teasingemo_Teasing0.142+0.260.159+0.290.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.058+0.020.014-0.040.0410.697
39Triumphemo_Triumph0.033+0.080.005+0.02-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg3.499+0.423.092+0.092.9841.238
41Arousal (AROU)vn_AROU_reg2.773+0.362.686+0.292.3591.147
42Arousal shift (ARSH)vn_ARSH_reg3.166+0.682.952+0.432.5800.867
43Vocal attack (ATCK)vn_ATCK_reg2.901+0.572.571+0.322.1621.295
44Background noise (BKGN)vn_BKGN_reg2.736+0.042.697-0.012.7030.761
45Brightness (BRGT)vn_BRGT_reg1.838-0.191.669-0.401.9910.813
46Speech chunking (CHNK)vn_CHNK_reg3.255+1.143.226+1.111.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg2.281+0.282.089+0.082.0180.945
48Cognitive load (COGL)vn_COGL_reg2.341+0.232.253+0.152.0941.061
49Dynamic arc (DARC)vn_DARC_reg1.825+0.151.941+0.241.6151.380
50Disfluency (DFLU)vn_DFLU_reg2.027-0.391.911-0.492.4881.184
51Emphasis (EMPH)vn_EMPH_reg3.154+0.363.154+0.362.6071.524
52Esthetics (ESTH)vn_ESTH_reg2.438+0.042.317-0.112.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.502+1.290.528+1.370.0860.322
54Focus (FOCS)vn_FOCS_reg2.727+0.352.500+0.142.3441.102
55Fullness (FULL)vn_FULL_reg1.126-0.731.140-0.721.7780.889
56Perceived gender (GEND)vn_GEND_reg4.790+0.464.727+0.424.0551.588
57Harmonicity (HARM)vn_HARM_reg2.405-0.092.499+0.022.4800.836
58Metallic character (METL)vn_METL_reg2.559+0.982.447+0.861.6210.958
59Pitch range (RANG)vn_RANG_reg2.616-0.332.453-0.512.9140.897
60Recording quality (RCQL)vn_RCQL_reg2.280-0.532.368-0.432.7540.888
61Vocal register (REGS)vn_REGS_reg0.157-0.820.225-0.771.1981.268
62Respiration (RESP)vn_RESP_reg1.965-0.681.877-0.762.6801.056
63Roughness (ROUG)vn_ROUG_reg2.836+0.902.836+0.902.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg2.870+0.872.671+0.642.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg2.125-0.032.113-0.042.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg1.558-0.351.512-0.391.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg1.929+0.591.863+0.511.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg3.683+0.853.588+0.762.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg2.271+0.122.179-0.012.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg2.584+0.642.408+0.422.0700.803
71Smoothness (SMTH)vn_SMTH_reg2.757+0.562.835+0.632.1431.093
72Stance (STNC)vn_STNC_reg3.641+0.893.464+0.732.6331.137
73Speech structure (STRU)vn_STRU_reg2.884+0.783.061+0.962.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg-0.083-1.03-0.158-1.081.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg2.892+1.732.932+1.771.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg1.796+0.172.019+0.331.5601.401
77Casual style (S_CASU)vn_S_CASU_reg3.861+0.053.763-0.023.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg1.687-0.691.340-0.952.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg3.036+0.272.856+0.122.7211.148
80Formal style (S_FORM)vn_S_FORM_reg1.809+0.602.073+0.861.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg3.243+0.333.014+0.152.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg2.282+0.672.579+0.881.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg1.805+2.462.140+3.050.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg2.437+0.072.275-0.062.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg2.331+0.871.828+0.481.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg3.632+0.843.360+0.642.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg-0.765-1.15-0.566-0.990.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg1.061-0.330.755-0.521.5971.618
89Tempo (TEMP)vn_TEMP_reg3.428+1.103.372+1.052.1171.190
90Tension (TENS)vn_TENS_reg1.871+0.371.877+0.371.5590.852
91Valence (VALN)vn_VALN_reg2.698+0.332.426+0.092.3281.118
92Valence shift (VALS)vn_VALS_reg3.034+0.482.654+0.132.5181.079
93Velocity flux (VFLX)vn_VFLX_reg2.452+0.162.450+0.162.2561.208
94Volatility (VOLT)vn_VOLT_reg1.826-0.861.667-1.042.5860.886
95Vulnerability (VULN)vn_VULN_reg1.069-1.611.107-1.572.6480.984
96Warmth (WARM)vn_WARM_reg2.570-0.012.181-0.442.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_62.402+0.041.853-0.302.3321.621
98Vocal-burst blend (0–10)blend_0_104.287-0.873.862-1.036.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.840-0.750.705-1.161.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.313+0.213.356+0.363.2500.293
101Arousaleiv_extra:Arousal1.528+0.551.644+0.821.2890.435
102Authenticityeiv_extra:Authenticity2.742+0.492.698+0.202.6670.153
103Background Noiseeiv_extra:Background_Noise0.453-1.340.272-2.030.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.154+0.381.107+0.221.0400.303
105Gendereiv_extra:Gender-1.106-0.49-1.130-0.51-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched2.153+1.302.171+1.371.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive2.028+0.362.060+0.431.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.433+0.872.486+1.032.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous0.865-0.191.002+0.210.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.129-0.480.192-0.280.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.803+1.320.914+1.660.3760.324
112Valenceeiv_extra:Valence-0.077-0.220.026+0.060.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached2.074+0.822.144+0.981.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold-0.034-0.280.087+0.120.0510.300
115score background qualityeiv_extra:score_background_quality3.602+0.433.675+0.543.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.694+0.624.809+1.004.5070.303
117score overall qualityeiv_extra:score_overall_quality2.933+0.552.973+0.632.6520.507
118score speech qualityeiv_extra:score_speech_quality1.975+1.151.995+1.281.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.526+0.715.776+0.964.7991.018
120Content usefulness (CU)audiobox:CU6.226+0.586.044+0.405.6491.001
121Production complexity (PC)audiobox:PC1.498-0.511.505-0.512.3311.617
122Production quality (PQ)audiobox:PQ7.289+0.287.117+0.087.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw4.061+0.244.033+0.203.9000.669
124Background quality, rawdnsmos:BAK_raw4.043+0.744.113+0.803.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.643+0.693.669+0.723.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.585+0.193.565+0.153.4880.501
127Background quality, calibrateddnsmos:BAK3.919+0.573.988+0.653.3890.928
128Overall quality, calibrateddnsmos:OVRL3.207+0.613.224+0.632.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.837+0.843.799+0.763.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.005+0.01-0.008-0.010.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.671-0.500.494-0.820.9470.554
132Agevoiceclap_attribute:Age3.248+0.043.290+0.113.2290.547
133Amusementvoiceclap_attribute:Amusement0.544+0.270.427+0.060.3920.572
134Angervoiceclap_attribute:Anger1.635+1.271.588+1.180.9920.506
135Arousalvoiceclap_attribute:Arousal2.075+1.052.077+1.061.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.511+0.010.714+0.430.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.786+0.102.705-0.392.7700.167
138Awevoiceclap_attribute:Awe0.761+0.860.757+0.840.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.421-0.550.501-0.250.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.840+1.050.762+0.720.5920.236
141Concentrationvoiceclap_attribute:Concentration1.245+0.061.453+0.451.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.075+0.010.957-0.241.0690.468
143Confusionvoiceclap_attribute:Confusion0.612-0.090.501-0.430.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.610-0.420.530-0.620.7800.406
145Contemptvoiceclap_attribute:Contempt1.290+1.721.118+1.210.7090.339
146Contentmentvoiceclap_attribute:Contentment0.436-0.500.432-0.510.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.977+0.751.048+0.990.7600.291
148Disgustvoiceclap_attribute:Disgust0.937+1.510.953+1.580.6250.207
149Distressvoiceclap_attribute:Distress1.182+0.731.161+0.670.9270.350
150Doubtvoiceclap_attribute:Doubt0.318-1.150.287-1.320.5350.188
151Elationvoiceclap_attribute:Elation0.795+0.640.695+0.390.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.216-0.290.060-0.790.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.627-0.140.667+0.010.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.421-0.160.340-0.470.4620.261
155Fearvoiceclap_attribute:Fear0.827+0.990.833+1.010.6180.212
156Gendervoiceclap_attribute:Gender-1.501-1.34-1.619-1.46-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.647-0.040.729+0.260.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched2.324+1.292.406+1.511.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism1.056+0.891.000+0.750.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability1.810+1.402.001+1.751.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.605-0.010.573-0.110.6080.332
162Interestvoiceclap_attribute:Interest1.770+0.391.594-0.011.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.590+0.190.802+0.650.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.813+1.180.848+1.330.5230.245
165Longingvoiceclap_attribute:Longing0.591+0.240.549+0.050.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice1.252+1.211.184+1.020.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive2.016+0.411.885+0.191.7750.594
168Painvoiceclap_attribute:Pain0.406+0.110.486+0.480.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.565+0.210.396-0.360.5030.301
170Pridevoiceclap_attribute:Pride1.007+0.651.049+0.770.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.368+0.172.314+0.002.3130.328
172Reliefvoiceclap_attribute:Relief0.945-0.190.842-0.441.0250.415
173Sadnessvoiceclap_attribute:Sadness0.802+0.160.841+0.280.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.679-0.220.552-0.440.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.882+0.590.934+0.710.6350.419
176Shamevoiceclap_attribute:Shame0.589+0.220.685+0.560.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh-0.330-1.36-0.555-1.980.1600.360
178Sournessvoiceclap_attribute:Sourness1.104+1.371.061+1.220.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant1.201+1.611.344+2.010.6230.359
180Teasingvoiceclap_attribute:Teasing0.430-0.290.247-0.780.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.789+0.761.764+0.721.3840.529
182Triumphvoiceclap_attribute:Triumph0.949+0.651.044+0.910.7160.359
183Valencevoiceclap_attribute:Valence-1.017-1.59-1.225-1.92-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached2.222+1.052.387+1.351.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold-0.402-0.84-0.659-1.48-0.0660.400
186durationvoiceclap_attribute:duration8.345-0.409.396-0.2410.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.453+0.423.638+0.793.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.748+0.624.795+0.954.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.795+0.392.903+0.652.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.979+1.271.988+1.411.8910.069
191talking speedvoiceclap_attribute:talking_speed20.973+3.1821.735+3.389.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
0-0.00530.0095-0.0006
10.05100.0151-0.0880
2-0.0665-0.0690-0.0868
3-0.0245-0.0849-0.0897
4-0.0828-0.1156-0.0832
5-0.1336-0.1074-0.1120
6-0.0465-0.0308-0.0802
70.10160.07830.1226
80.0041-0.03200.0806
9-0.1016-0.1025-0.0501
100.06150.04490.0770
11-0.0903-0.1079-0.0530
12-0.0585-0.0229-0.0491
13-0.0930-0.0877-0.0873
14-0.0812-0.1030-0.0922
15-0.0775-0.0811-0.0793
16-0.1096-0.0707-0.0521
17-0.1165-0.0959-0.0763
18-0.0604-0.0219-0.0586
19-0.0577-0.0707-0.0837
200.18710.15120.1221
21-0.03090.0112-0.0290
22-0.0893-0.0844-0.0833
23-0.0226-0.03810.0846
24-0.0877-0.0838-0.0714
25-0.0013-0.0575-0.0353
260.00080.0107-0.0765
27-0.0526-0.02000.0544
28-0.0420-0.0405-0.0170
29-0.00370.05150.0568
300.12770.18960.1793
310.04840.0570-0.0394
32-0.1058-0.0915-0.0853
33-0.1219-0.1134-0.0975
34-0.1053-0.1161-0.0793
350.20100.26190.2231
36-0.0855-0.1041-0.0951
37-0.1133-0.0920-0.1048
38-0.0484-0.0762-0.0668
39-0.0930-0.0992-0.0893
40-0.0727-0.0573-0.0367
41-0.1293-0.1019-0.1272
420.07750.07720.0728
430.08710.05620.0746
440.16890.24760.1925
450.16250.16980.1665
46-0.0941-0.0542-0.0345
47-0.0973-0.0915-0.0895
48-0.0183-0.0597-0.0864
490.06520.05920.0450
500.03630.04160.0361
51-0.0327-0.0600-0.0885
52-0.0187-0.01860.0176
530.08390.10410.0886
54-0.1251-0.0871-0.0991
550.24590.18960.1434
56-0.0126-0.0194-0.0839
57-0.1144-0.1479-0.0970
580.03880.01290.0584
590.11760.10630.1075
60-0.0668-0.0816-0.0853
610.33990.24870.2576
62-0.0748-0.0904-0.0828
63-0.0334-0.0575-0.0583
64-0.00600.04930.0522
65-0.0909-0.0937-0.0722
660.07910.09920.0529
67-0.0834-0.0751-0.0853
68-0.0727-0.1068-0.0936
69-0.0791-0.0866-0.0811
700.05690.04110.0436
71-0.0146-0.0051-0.0109
72-0.0684-0.0641-0.0856
730.02950.02620.0368
74-0.0941-0.0751-0.0967
75-0.00760.00030.0810
760.03530.01680.0243
77-0.0460-0.0292-0.0808
780.05240.0264-0.0215
79-0.0572-0.0767-0.0768
80-0.03170.0263-0.0481
810.09090.08440.0305
82-0.0802-0.1052-0.0847
83-0.0882-0.1068-0.0906
840.08770.12440.1295
85-0.00920.04960.0451
86-0.0681-0.0696-0.0632
87-0.0914-0.0668-0.0551
880.03130.03050.0292
89-0.0227-0.0221-0.0812
90-0.0791-0.1052-0.0959
910.18600.13150.1504
92-0.1090-0.0937-0.0859
930.09190.08440.1229
940.09570.08990.1160
95-0.0521-0.0197-0.0535
96-0.0770-0.0756-0.0510
970.0031-0.00590.0567
980.08550.07890.1327
990.0125-0.02220.0599
1000.08390.13640.0795
1010.11760.10900.0682
102-0.0473-0.0499-0.0565
103-0.1005-0.0937-0.0934
104-0.0665-0.0603-0.0763
105-0.0492-0.0690-0.0236
106-0.0251-0.0425-0.0548
1070.03340.0284-0.0204
108-0.00420.0064-0.0374
109-0.0577-0.0866-0.0868
1100.0015-0.0132-0.0205
111-0.0919-0.0904-0.1027
112-0.0855-0.0986-0.0872
1130.08230.07510.0681
114-0.0732-0.0616-0.0551
1150.07110.08490.1692
1160.04780.0208-0.0102
1170.07640.12490.1534
118-0.1358-0.1205-0.0862
1190.12030.15120.0608
120-0.0834-0.0679-0.0910
1210.06250.05340.1449
122-0.0951-0.0663-0.0909
123-0.0575-0.0161-0.0329
124-0.0295-0.0241-0.0872
1250.01220.03620.0454
126-0.0310-0.0644-0.0032
1270.01260.00860.0963
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
00.03600.03100.1121
10.00280.00330.0621
20.01570.00640.0616
3-0.0151-0.0070-0.0348
4-0.03130.0208-0.0427
5-0.1054-0.0713-0.0490
6-0.01450.05100.0202
70.04420.0358-0.0475
80.00190.05950.0087
90.04420.00680.0229
100.07720.08320.0803
11-0.0388-0.06130.0465
12-0.0459-0.0722-0.0513
13-0.0408-0.0179-0.0414
14-0.0564-0.04310.0609
150.05540.07770.1038
160.09790.08860.0659
170.0023-0.00420.0259
18-0.0918-0.0372-0.0477
19-0.0218-0.0172-0.0052
200.01420.03430.0117
21-0.0561-0.0650-0.0435
22-0.0202-0.0367-0.0486
23-0.0945-0.0722-0.0453
240.02870.04490.0153
25-0.02870.0616-0.0804
260.01610.0083-0.0251
270.03090.05370.1073
280.02120.0200-0.0079
29-0.0445-0.0589-0.0451
300.10270.06280.0035
310.11420.09770.1494
32-0.01960.0156-0.0482
330.11010.06620.0439
340.0003-0.0012-0.0476
350.0014-0.0001-0.0440
36-0.0010-0.03730.0139
370.02750.0337-0.0324
380.05130.0656-0.0036
39-0.0578-0.0431-0.0196
40-0.0615-0.0692-0.0543
410.07990.05620.0492
42-0.0260-0.0231-0.0466
430.01910.0771-0.0554
440.05510.0722-0.0050
45-0.0541-0.00540.0227
46-0.0272-0.0665-0.0287
47-0.0326-0.0467-0.0499
480.0070-0.0184-0.0500
49-0.0435-0.0707-0.0460
50-0.0272-0.0212-0.0649
510.11150.08560.0244
520.03130.0540-0.0300
53-0.01690.00430.0180
540.02960.09290.0596
550.04760.05040.0651
560.05300.08380.1099
570.0136-0.0214-0.0429
580.00000.01500.0061
590.02480.04490.0641
600.01780.04400.0110
61-0.0476-0.0510-0.0435
620.0036-0.0217-0.0412
63-0.0127-0.0137-0.0510
640.0022-0.0410-0.0450
650.05440.05770.0431
660.0076-0.0331-0.0591
670.0243-0.0106-0.0493
68-0.0490-0.0474-0.0505
690.03080.0826-0.0194
70-0.0643-0.0710-0.0680
71-0.01060.0061-0.0059
72-0.0405-0.0070-0.0265
730.09720.0501-0.0028
74-0.01950.0077-0.0506
750.11420.11110.0797
76-0.01730.0410-0.0140
77-0.0816-0.0604-0.0274
78-0.0109-0.04130.0009
79-0.00290.01870.0190
80-0.1081-0.1069-0.0498
81-0.00590.0110-0.0562
82-0.0308-0.0443-0.0450
83-0.0670-0.0477-0.0416
840.04620.03080.1108
850.03470.04430.0987
86-0.0442-0.0062-0.0447
87-0.0238-0.01780.1187
880.07950.08010.0365
890.03110.0073-0.0453
90-0.0466-0.0299-0.0496
91-0.0925-0.0844-0.0589
920.0016-0.0054-0.0120
930.01650.0595-0.0530
94-0.0500-0.0543-0.0753
95-0.0551-0.0320-0.0522
96-0.00910.03430.0045
97-0.0571-0.0586-0.0590
980.02140.0431-0.0529
99-0.02020.03870.0098
100-0.0396-0.0389-0.0308
101-0.0350-0.0149-0.0473
1020.03010.0231-0.0096
103-0.00400.03450.0292
104-0.1149-0.1050-0.0652
105-0.0433-0.05460.0252
106-0.0823-0.0653-0.0386
107-0.0772-0.0437-0.0449
1080.19450.21490.2477
109-0.01780.01440.0565
110-0.0792-0.0698-0.0596
111-0.1163-0.0677-0.0413
112-0.0449-0.0552-0.0404
1130.07850.07070.0152
114-0.0274-0.0091-0.0225
1150.02750.0363-0.0281
116-0.0311-0.05010.0144
1170.0064-0.0244-0.0629
118-0.0860-0.0513-0.0123
119-0.0326-0.0340-0.0515
1200.0031-0.0322-0.0350
121-0.0466-0.0285-0.0538
122-0.0809-0.0929-0.0473
1230.04830.02640.1212
1240.04760.0128-0.0171
125-0.0986-0.1062-0.0487
1260.11420.12990.2008
127-0.0184-0.0250-0.0515
1280.00410.0034-0.0201
1290.10740.10320.1058
130-0.0282-0.0583-0.0503
1310.04760.0232-0.0435
132-0.01730.0410-0.0251
1330.08120.03660.1243
1340.05170.04370.0644
135-0.0115-0.0063-0.0504
136-0.0371-0.0543-0.0299
1370.06870.02230.0356
1380.00820.0323-0.0253
1390.03960.0574-0.0512
140-0.0277-0.0165-0.0092
141-0.00620.0413-0.0219
142-0.0598-0.0322-0.0433
1430.04350.02160.0036
144-0.1074-0.1014-0.0522
1450.00520.0074-0.0023
1460.13050.16390.0696
147-0.0087-0.01970.0011
1480.09310.08560.0641
1490.12780.13110.1825
1500.12850.09840.0220
1510.01340.01800.0264
152-0.0108-0.0098-0.0555
1530.08770.02960.0548
1540.02310.02110.0120
1550.0711-0.0182-0.0151
1560.0121-0.0036-0.0170
1570.05980.06250.0134
158-0.0001-0.0296-0.0623
1590.04060.06710.0080
1600.07040.07650.1307
1610.0112-0.0149-0.0703
162-0.0840-0.0461-0.0324
163-0.0918-0.0668-0.0557
164-0.0999-0.0386-0.0440
1650.02070.03080.0189
1660.01070.0556-0.0392
167-0.1537-0.1366-0.0464
1680.09520.09230.0891
1690.09450.0522-0.0511
170-0.0051-0.0276-0.0469
171-0.0331-0.0540-0.0440
1720.09930.11900.0351
1730.02140.06590.1326
174-0.0636-0.0419-0.0514
175-0.0775-0.0856-0.0829
1760.05170.03890.1128
177-0.0352-0.0269-0.0413
1780.04160.0304-0.0439
1790.11150.12320.0486
1800.17130.18940.1168
181-0.01330.00380.0148
182-0.0578-0.0683-0.0432
183-0.1027-0.0768-0.0767
1840.02600.07590.0411
1850.06460.06370.0135
186-0.0731-0.0777-0.0604
1870.10270.12510.0756
1880.07210.05040.0840
1890.05810.06400.0453
190-0.04200.00730.0312
1910.05100.04460.0766
1920.09180.13540.0782
1930.04420.06500.0864
1940.00620.01680.0020
1950.01310.0217-0.0014
1960.00270.02000.0414
1970.13870.12020.1052
198-0.0008-0.0165-0.0484
1990.03180.0407-0.0339
200-0.0140-0.0205-0.0579
2010.11010.13540.0426
202-0.0643-0.0838-0.1041
2030.12650.09230.0235
2040.11010.07040.1121
205-0.0064-0.01030.0591
2060.02060.03260.0404
207-0.02620.01680.0458
208-0.0189-0.0619-0.0519
209-0.00600.0067-0.0122
2100.11490.08740.1063
2110.0102-0.02430.0223
212-0.00640.06950.1444
2130.03620.02490.0977
2140.25970.29260.2761
215-0.0362-0.0370-0.0421
216-0.0287-0.0501-0.0599
217-0.0537-0.02820.0110
218-0.0030-0.04830.0520
219-0.02110.0087-0.0481
220-0.0595-0.0789-0.0553
221-0.0071-0.0083-0.0158
2220.00880.02630.0431
223-0.0602-0.0381-0.0401
224-0.02130.00730.0799
2250.02920.02410.0587
226-0.0316-0.01190.0065
2270.05920.11780.0442
2280.00390.0125-0.0696
2290.05920.0098-0.0549
2300.03250.05160.0892
2310.06260.03980.0690
232-0.01680.0023-0.0005
233-0.0350-0.0577-0.0809
2340.16660.12570.0206
2350.15770.13110.0696
236-0.0001-0.0114-0.0430
237-0.01200.0059-0.0511
2380.06360.03600.1029
239-0.0139-0.02110.0218
2400.07170.10200.1371
2410.00540.02880.0710
242-0.0325-0.0384-0.0737
243-0.0860-0.0971-0.0643
244-0.0690-0.0695-0.0532
245-0.0843-0.0571-0.0476
246-0.01650.0610-0.0405
247-0.0728-0.0923-0.0421
248-0.0724-0.0744-0.0572
249-0.0609-0.0392-0.0345
Source: HUMANENESS_SMALL_VOICE_IDENTITY_R1_R2_2026-10-02_mediathek_0047.html · SHA-256 8abb604125492b46…
g06EnglishMediathek voice study

mediathek_0047 · en · 13/50

Generated TTS prediction · Rank 1 qv · 12.0 s

For folks constantly on the move, trying to gather all that intel about a search subject by hand is just too much work. (normal breathing) That's where online tools like People Finder really shine, you know?

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Impatience and Irritability
z +2.16 / +1.66raw 1.67 / 1.32
Hope Enthusiasm Optimism
z +1.07 / +1.74raw 0.93 / 1.44
Anger
z +1.56 / +1.42raw 0.96 / 0.87
Sourness
z +1.32 / +1.34raw 0.57 / 0.58
Contemplation
z +1.03 / +1.33raw 0.70 / 0.88
Pride
z +0.59 / +1.27raw 0.30 / 0.65
Concentration
z +1.13 / +1.17raw 1.47 / 1.50

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Authoritative style (S_AUTH)
z +1.95 / +1.75raw 3.12 / 2.92
Ranting or angry style (S_RANT)
z +1.84 / +1.77raw 3.59 / 3.50
Stance (STNC)
z +1.82 / +1.81raw 4.70 / 4.69
Nasal resonance (R_NASL)
z +1.75 / +1.78raw 4.62 / 4.66
Tempo (TEMP)
z +1.50 / +1.61raw 3.91 / 4.03
Roughness (ROUG)
z +1.31 / +1.53raw 3.19 / 3.39
Tension (TENS)
z +1.48 / +1.39raw 2.82 / 2.75

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z +0.55 / +0.44raw 3.22 / 3.05
Vocal-burst blend (0–10)z +0.05 / +0.10raw 6.82 / 6.96
Content enjoyment (CE)z +0.64 / +0.47raw 5.45 / 5.28
Production quality (PQ)z +0.32 / +0.23raw 7.33 / 7.25
Speech signal quality, calibratedz +0.14 / +0.19raw 3.56 / 3.59
Background quality, calibratedz +0.69 / +0.63raw 4.03 / 3.98
Overall quality, calibratedz +0.73 / +0.65raw 3.29 / 3.24
P.808 MOS estimatez +0.55 / +0.51raw 3.70 / 3.67
score content enjoymentz +0.74 / +1.10raw 4.73 / 4.84
score overall qualityz +0.86 / +0.91raw 3.09 / 3.11
Background Noisez -0.41 / -0.98raw 0.70 / 0.55
Tempo (TEMP)z +1.50 / +1.61raw 3.91 / 4.03
Arousal (AROU)z +1.45 / +1.39raw 4.02 / 3.95
Valence (VALN)z +0.52 / +0.41raw 2.91 / 2.78
Recording quality (RCQL)z -0.76 / -0.91raw 2.08 / 1.95
Background noise (BKGN)z -0.62 / -0.58raw 2.23 / 2.26
talking speedz +2.38 / +2.79raw 17.98 / 19.51
Characters per second (CPS): 17.32 raw · z +2.21

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.9070.886
Identity · 250D0.6830.673

Vocal-burst regions

Tiny
Base
0 s12.0 s

Verified burst intervals are unavailable for this clip.

Tiny: 0 regions, count-head estimate 0.00; Base: 0 regions, count-head estimate 0.00.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.094+0.150.215+0.370.0070.570
1Amusementemo_Amusement0.066+0.080.171+0.19-0.0010.885
2Angeremo_Anger0.959+1.560.873+1.420.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.036+0.050.029+0.040.0050.572
4Aweemo_Awe-0.002-0.01-0.012-0.030.0000.427
5Bitternessemo_Bitterness0.382+0.890.325+0.750.0130.415
6Concentrationemo_Concentration1.470+1.131.498+1.170.6340.741
7Confusionemo_Confusion0.067+0.120.011+0.020.0010.540
8Contemplationemo_Contemplation0.695+1.030.878+1.330.0820.598
9Contemptemo_Contempt0.365+0.790.496+1.07-0.0000.464
10Contentmentemo_Contentment0.235+0.460.527+1.030.0000.514
11Disappointmentemo_Disappointment0.571+0.820.260+0.370.0000.700
12Disgustemo_Disgust0.147+0.290.089+0.170.0070.489
13Distressemo_Distress0.224+0.270.061+0.070.0000.821
14Doubtemo_Doubt0.139+0.290.150+0.310.0060.466
15Elationemo_Elation0.243+0.320.372+0.490.0120.728
16Embarrassmentemo_Embarrassment-0.027-0.16-0.007-0.120.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.113+0.120.037+0.040.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.427+0.740.054+0.050.0280.539
19Fearemo_Fear-0.008-0.04-0.015-0.050.0150.610
20Helplessnessemo_Helplessness0.168+0.250.060+0.09-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.930+1.071.440+1.740.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability1.669+2.161.321+1.660.1490.705
23Infatuationemo_Infatuation-0.016-0.000.003+0.03-0.0150.533
24Interestemo_Interest2.149+0.622.392+0.891.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness-0.040-0.03-0.022+0.01-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.377+0.720.419+0.810.0590.442
27Longingemo_Longing0.158+0.110.035-0.080.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.057+0.000.092+0.070.0550.534
29Painemo_Pain-0.029-0.05-0.031-0.05-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.018+0.000.143+0.210.0170.614
31Prideemo_Pride0.301+0.590.647+1.270.0020.509
32Reliefemo_Relief0.040+0.010.035+0.000.0330.681
33Sadnessemo_Sadness0.110+0.14-0.037-0.050.0000.759
34Sexual Lustemo_Sexual_Lust-0.033-0.130.018-0.020.0300.494
35Shameemo_Shame0.005-0.010.010-0.010.0130.510
36Sournessemo_Sourness0.575+1.320.582+1.34-0.0050.439
37Teasingemo_Teasing0.124+0.230.205+0.380.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.146+0.150.362+0.460.0410.697
39Triumphemo_Triumph0.015+0.040.088+0.18-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg3.341+0.293.031+0.042.9841.238
41Arousal (AROU)vn_AROU_reg4.021+1.453.950+1.392.3591.147
42Arousal shift (ARSH)vn_ARSH_reg3.690+1.283.717+1.312.5800.867
43Vocal attack (ATCK)vn_ATCK_reg3.363+0.933.343+0.912.1621.295
44Background noise (BKGN)vn_BKGN_reg2.232-0.622.265-0.582.7030.761
45Brightness (BRGT)vn_BRGT_reg2.733+0.912.812+1.011.9910.813
46Speech chunking (CHNK)vn_CHNK_reg3.313+1.183.167+1.071.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg2.983+1.023.103+1.152.0180.945
48Cognitive load (COGL)vn_COGL_reg1.702-0.371.579-0.492.0941.061
49Dynamic arc (DARC)vn_DARC_reg3.368+1.273.379+1.281.6151.380
50Disfluency (DFLU)vn_DFLU_reg2.725+0.202.718+0.192.4881.184
51Emphasis (EMPH)vn_EMPH_reg4.289+1.104.350+1.142.6071.524
52Esthetics (ESTH)vn_ESTH_reg2.169-0.302.158-0.312.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.419+1.030.456+1.150.0860.322
54Focus (FOCS)vn_FOCS_reg3.464+1.023.507+1.062.3441.102
55Fullness (FULL)vn_FULL_reg1.078-0.791.029-0.841.7780.889
56Perceived gender (GEND)vn_GEND_reg4.808+0.474.653+0.384.0551.588
57Harmonicity (HARM)vn_HARM_reg2.665+0.222.779+0.362.4800.836
58Metallic character (METL)vn_METL_reg2.642+1.062.443+0.861.6210.958
59Pitch range (RANG)vn_RANG_reg3.588+0.753.483+0.632.9140.897
60Recording quality (RCQL)vn_RCQL_reg2.080-0.761.950-0.912.7540.888
61Vocal register (REGS)vn_REGS_reg0.844-0.280.883-0.251.1981.268
62Respiration (RESP)vn_RESP_reg3.248+0.543.440+0.722.6801.056
63Roughness (ROUG)vn_ROUG_reg3.194+1.313.387+1.532.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg2.728+0.702.479+0.412.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg2.459+0.382.558+0.502.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg2.235+0.322.411+0.501.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg1.144-0.351.053-0.461.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg4.624+1.754.657+1.782.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg2.992+1.093.027+1.142.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg1.888-0.231.669-0.502.0700.803
71Smoothness (SMTH)vn_SMTH_reg2.623+0.442.662+0.482.1431.093
72Stance (STNC)vn_STNC_reg4.700+1.824.691+1.812.6331.137
73Speech structure (STRU)vn_STRU_reg2.391+0.272.434+0.312.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg0.187-0.840.036-0.951.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg3.119+1.952.916+1.751.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg2.952+0.993.006+1.031.5601.401
77Casual style (S_CASU)vn_S_CASU_reg4.717+0.724.882+0.853.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg1.947-0.491.860-0.562.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg4.284+1.364.070+1.172.7211.148
80Formal style (S_FORM)vn_S_FORM_reg0.769-0.420.816-0.371.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg2.824+0.003.121+0.232.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg0.499-0.560.471-0.581.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.529+0.210.494+0.150.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg3.335+0.753.157+0.612.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg3.587+1.843.498+1.771.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg3.823+0.983.725+0.912.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg0.511-0.110.763+0.090.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg0.187-0.870.462-0.701.5971.618
89Tempo (TEMP)vn_TEMP_reg3.907+1.504.029+1.612.1171.190
90Tension (TENS)vn_TENS_reg2.817+1.482.745+1.391.5590.852
91Valence (VALN)vn_VALN_reg2.906+0.522.782+0.412.3281.118
92Valence shift (VALS)vn_VALS_reg2.558+0.042.385-0.122.5181.079
93Velocity flux (VFLX)vn_VFLX_reg2.839+0.482.891+0.532.2561.208
94Volatility (VOLT)vn_VOLT_reg2.889+0.342.874+0.322.5860.886
95Vulnerability (VULN)vn_VULN_reg1.107-1.571.138-1.532.6480.984
96Warmth (WARM)vn_WARM_reg2.222-0.392.086-0.542.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_63.216+0.553.051+0.442.3321.621
98Vocal-burst blend (0–10)blend_0_106.817+0.056.961+0.106.6762.738
99Legacy R_quality (no P3 test labels)R_quality1.222+0.381.206+0.331.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.317+0.233.225-0.093.2500.293
101Arousaleiv_extra:Arousal1.762+1.091.872+1.341.2890.435
102Authenticityeiv_extra:Authenticity2.726+0.382.746+0.522.6670.153
103Background Noiseeiv_extra:Background_Noise0.698-0.410.549-0.980.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.211+0.561.110+0.231.0400.303
105Gendereiv_extra:Gender-1.048-0.43-0.947-0.33-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched2.067+0.962.039+0.851.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive2.414+1.182.440+1.241.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.165+0.062.277+0.402.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous0.955+0.081.053+0.360.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.011-0.840.232-0.150.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant1.010+1.960.836+1.420.3760.324
112Valenceeiv_extra:Valence-0.155-0.42-0.011-0.040.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.819+0.241.752+0.091.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold-0.177-0.760.119+0.230.0510.300
115score background qualityeiv_extra:score_background_quality3.678+0.543.740+0.633.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.732+0.744.840+1.104.5070.303
117score overall qualityeiv_extra:score_overall_quality3.088+0.863.111+0.912.6520.507
118score speech qualityeiv_extra:score_speech_quality1.945+0.941.962+1.061.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.447+0.645.276+0.474.7991.018
120Content usefulness (CU)audiobox:CU6.058+0.416.128+0.485.6491.001
121Production complexity (PC)audiobox:PC1.972-0.221.744-0.362.3311.617
122Production quality (PQ)audiobox:PQ7.328+0.327.248+0.237.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw4.048+0.224.053+0.233.9000.669
124Background quality, rawdnsmos:BAK_raw4.250+0.924.149+0.833.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.770+0.843.691+0.753.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.560+0.143.585+0.193.4880.501
127Background quality, calibrateddnsmos:BAK4.028+0.693.977+0.633.3890.928
128Overall quality, calibrateddnsmos:OVRL3.285+0.733.236+0.652.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.696+0.553.674+0.513.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.002+0.000.002+0.000.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection1.060+0.201.086+0.250.9470.554
132Agevoiceclap_attribute:Age3.357+0.243.306+0.143.2290.547
133Amusementvoiceclap_attribute:Amusement0.798+0.710.701+0.540.3920.572
134Angervoiceclap_attribute:Anger2.246+2.482.198+2.380.9920.506
135Arousalvoiceclap_attribute:Arousal2.666+2.002.680+2.031.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.656+0.310.967+0.960.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.921+0.902.943+1.042.7700.167
138Awevoiceclap_attribute:Awe0.580+0.160.780+0.930.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.637+0.260.626+0.220.5670.264
140Bitternessvoiceclap_attribute:Bitterness1.061+1.991.044+1.910.5920.236
141Concentrationvoiceclap_attribute:Concentration1.012-0.381.312+0.181.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.079+0.021.263+0.411.0690.468
143Confusionvoiceclap_attribute:Confusion0.640-0.010.836+0.580.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.447-0.820.493-0.710.7800.406
145Contemptvoiceclap_attribute:Contempt1.378+1.981.314+1.790.7090.339
146Contentmentvoiceclap_attribute:Contentment0.543-0.150.464-0.410.5880.307
147Disappointmentvoiceclap_attribute:Disappointment1.369+2.091.446+2.360.7600.291
148Disgustvoiceclap_attribute:Disgust1.072+2.161.031+1.960.6250.207
149Distressvoiceclap_attribute:Distress1.481+1.591.615+1.970.9270.350
150Doubtvoiceclap_attribute:Doubt0.338-1.050.455-0.420.5350.188
151Elationvoiceclap_attribute:Elation1.034+1.250.947+1.030.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.221-0.270.361+0.170.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.405-0.970.427-0.890.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.543+0.310.449-0.050.4620.261
155Fearvoiceclap_attribute:Fear0.807+0.890.834+1.020.6180.212
156Gendervoiceclap_attribute:Gender-1.414-1.26-1.312-1.16-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.778+0.450.954+1.100.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched2.266+1.142.284+1.181.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism1.451+1.881.519+2.060.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability2.518+2.712.848+3.321.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.352-0.770.543-0.190.6080.332
162Interestvoiceclap_attribute:Interest1.799+0.461.806+0.471.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.717+0.460.921+0.900.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.912+1.591.160+2.600.5230.245
165Longingvoiceclap_attribute:Longing0.516-0.100.525-0.060.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice1.479+1.871.525+2.000.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive2.846+1.802.921+1.931.7750.594
168Painvoiceclap_attribute:Pain0.441+0.270.633+1.150.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.734+0.770.692+0.630.5030.301
170Pridevoiceclap_attribute:Pride1.284+1.441.224+1.270.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.185-0.392.091-0.682.3130.328
172Reliefvoiceclap_attribute:Relief0.854-0.410.793-0.561.0250.415
173Sadnessvoiceclap_attribute:Sadness0.906+0.490.997+0.780.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.790-0.020.783-0.040.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.464-0.410.718+0.200.6350.419
176Shamevoiceclap_attribute:Shame0.794+0.940.929+1.420.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh-0.543-1.95-0.600-2.110.1600.360
178Sournessvoiceclap_attribute:Sourness1.186+1.651.225+1.790.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant1.613+2.761.372+2.090.6230.359
180Teasingvoiceclap_attribute:Teasing0.638+0.260.336-0.540.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude2.086+1.332.074+1.301.3840.529
182Triumphvoiceclap_attribute:Triumph1.213+1.381.132+1.160.7160.359
183Valencevoiceclap_attribute:Valence-1.592-2.50-2.067-3.25-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached1.533-0.191.420-0.391.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold-0.618-1.38-0.774-1.77-0.0660.400
186durationvoiceclap_attribute:duration11.289+0.0611.646+0.1210.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.509+0.533.653+0.823.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.720+0.444.675+0.134.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.923+0.702.973+0.822.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.970+1.141.959+0.981.8910.069
191talking speedvoiceclap_attribute:talking_speed17.983+2.3819.507+2.799.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.04710.01980.0130
1-0.02340.0204-0.0855
2-0.0651-0.0565-0.0844
3-0.0471-0.1118-0.0872
4-0.1090-0.1182-0.0809
5-0.1303-0.1177-0.1089
6-0.0322-0.0229-0.0780
70.04280.09260.1024
80.0527-0.03190.0675
9-0.0719-0.0781-0.0079
100.06800.04720.0956
11-0.0556-0.0903-0.0572
12-0.0223-0.0192-0.0074
13-0.0892-0.0746-0.0849
14-0.0605-0.0897-0.0846
15-0.0906-0.0920-0.0771
16-0.0697-0.0390-0.0649
17-0.0690-0.1124-0.0406
18-0.0033-0.0213-0.0356
19-0.0811-0.0600-0.0814
200.12320.13160.1124
21-0.0026-0.0069-0.0238
22-0.0800-0.0743-0.0809
23-0.0434-0.05740.0479
24-0.0779-0.0652-0.0695
25-0.0418-0.0620-0.0623
26-0.01500.0130-0.0807
27-0.00850.00220.0948
28-0.0239-0.0513-0.0203
290.00320.11530.0784
300.22370.19100.1646
310.0025-0.0460-0.0720
32-0.0927-0.0868-0.0829
33-0.1048-0.0984-0.0948
34-0.0271-0.0571-0.0771
350.25200.31920.2593
36-0.0814-0.0856-0.0925
37-0.0843-0.1008-0.0808
38-0.0556-0.0786-0.0806
39-0.1041-0.1037-0.0868
40-0.0556-0.0571-0.0417
41-0.1416-0.1182-0.1237
420.05730.06030.0535
430.03270.04720.0161
440.30870.26210.2137
450.07540.12350.1271
46-0.1012-0.0804-0.0553
47-0.0867-0.0914-0.0870
48-0.0135-0.0373-0.0781
490.04500.03900.0533
500.06120.07980.0515
51-0.1062-0.0880-0.0860
520.04040.00440.0434
530.07470.07160.0800
54-0.1062-0.1194-0.0963
550.19120.21090.1543
56-0.0151-0.0469-0.0816
57-0.1175-0.1444-0.0944
580.01660.00670.0556
590.02800.03190.0718
60-0.0437-0.0454-0.0830
610.23930.21320.2498
62-0.0913-0.0626-0.0804
63-0.0250-0.0728-0.0817
640.03470.07280.0214
65-0.0935-0.0821-0.0702
660.11680.12230.0883
67-0.0977-0.0705-0.0830
68-0.1260-0.1002-0.0909
69-0.0835-0.0850-0.0789
700.10760.09960.0889
71-0.02070.0013-0.0190
72-0.0775-0.0533-0.0833
730.03890.05770.0536
74-0.1189-0.0702-0.0940
750.03660.05130.1032
760.00870.03470.0209
77-0.0221-0.0092-0.0694
780.0388-0.0161-0.0061
79-0.0811-0.0719-0.0746
80-0.0942-0.0218-0.0779
810.04640.0833-0.0023
82-0.0814-0.0996-0.0823
83-0.1133-0.1025-0.0881
840.06900.07920.1362
850.03350.04310.0212
86-0.0906-0.0880-0.0359
87-0.0789-0.0909-0.0541
88-0.0230-0.0530-0.0518
89-0.0619-0.0518-0.0790
90-0.1041-0.1182-0.0933
910.14940.12640.1476
92-0.1005-0.0647-0.0836
930.17910.14620.1476
940.08640.07460.0850
95-0.0023-0.0119-0.0616
96-0.1019-0.0632-0.0864
97-0.01020.00980.0358
980.04850.03650.0994
990.07150.00020.1017
1000.06870.10720.0753
1010.06340.10780.0679
102-0.0368-0.0513-0.0375
103-0.1069-0.0714-0.0908
104-0.0489-0.0536-0.0775
105-0.0605-0.07080.0001
106-0.0765-0.0664-0.1035
1070.01800.0396-0.0022
108-0.00040.0181-0.0317
109-0.0704-0.0850-0.0844
1100.0142-0.0316-0.0179
111-0.0761-0.0786-0.0999
112-0.0935-0.0845-0.0848
1130.14090.13400.1059
114-0.0448-0.03670.0175
1150.07820.09090.1707
116-0.0489-0.0052-0.0458
1170.11750.10250.1576
118-0.1196-0.1165-0.0838
1190.10900.13340.0360
120-0.0761-0.0661-0.0717
1210.05170.03490.1310
122-0.0726-0.0734-0.0884
123-0.1232-0.0454-0.0801
124-0.0036-0.0119-0.0724
1250.03220.02490.0513
126-0.0271-0.07430.0001
1270.0365-0.00500.1081
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
00.02500.02240.0592
1-0.0012-0.00970.0052
20.04740.02650.0562
3-0.02770.0137-0.0368
40.01050.0061-0.0422
5-0.0489-0.0624-0.0485
60.02220.0540-0.0089
70.00170.0117-0.0470
8-0.01150.0475-0.0105
90.0235-0.00180.0372
100.03150.02250.0583
11-0.0241-0.0550-0.0040
12-0.0543-0.0453-0.0508
13-0.0565-0.0369-0.0410
14-0.0481-0.0301-0.0085
150.03350.07700.1059
160.09260.09630.0428
17-0.0481-0.02440.0083
18-0.0609-0.0312-0.0314
190.0112-0.0040-0.0174
200.11960.07450.0646
21-0.0645-0.0708-0.0430
22-0.0343-0.0289-0.0482
23-0.1072-0.0863-0.0449
240.04390.05840.0143
250.00480.0447-0.0853
260.0653-0.01020.0039
270.03170.01750.0878
280.05030.07610.0321
29-0.0853-0.0869-0.0446
300.04410.0487-0.0237
310.13270.10990.1776
320.06090.0782-0.0478
330.03150.01250.0258
340.02520.0164-0.0475
35-0.00700.0070-0.0598
360.03830.00480.0440
370.02020.0077-0.0321
380.03590.0222-0.0344
39-0.0882-0.1012-0.0605
40-0.0467-0.0404-0.0537
410.06640.05740.0258
42-0.0284-0.0039-0.0461
430.07400.1273-0.0548
440.08570.09750.0192
45-0.0489-0.02550.0164
46-0.0463-0.0646-0.0275
47-0.00620.0067-0.0593
480.0034-0.0227-0.0495
49-0.0459-0.0602-0.0170
50-0.0456-0.0689-0.0867
510.04070.03820.0020
520.05400.0857-0.0066
530.01340.02110.0256
540.08820.11920.0942
550.04050.02510.0477
560.02720.04220.1199
57-0.0525-0.0711-0.0542
58-0.0277-0.0283-0.0207
590.01650.01330.0559
60-0.01550.0171-0.0124
61-0.0514-0.0612-0.0431
62-0.0407-0.0494-0.0408
63-0.0323-0.0290-0.0504
64-0.0225-0.0003-0.0517
650.02680.01930.0376
66-0.0292-0.0247-0.0585
67-0.0131-0.0167-0.0488
68-0.0478-0.0315-0.0500
690.00590.0696-0.0043
70-0.0872-0.0727-0.0673
710.00610.00250.0018
72-0.03350.0000-0.0185
730.12030.11180.0608
740.02150.0151-0.0501
750.06820.09130.0815
760.02610.05060.0094
77-0.0799-0.0612-0.0370
78-0.0283-0.04600.0085
790.05000.03510.0417
80-0.1437-0.1217-0.0493
810.01110.0238-0.0556
82-0.0496-0.0714-0.0446
83-0.0627-0.0259-0.0412
84-0.00090.04600.1010
850.00700.01390.0889
860.01650.0404-0.0443
87-0.0281-0.02450.0309
880.04650.04380.0684
890.02920.0258-0.0409
90-0.0780-0.0652-0.0491
91-0.0999-0.1466-0.0583
920.02480.02530.0226
930.05830.0543-0.0562
94-0.0441-0.0696-0.0660
95-0.0773-0.0621-0.0517
960.00440.02900.0353
97-0.0496-0.0362-0.0584
98-0.00150.0158-0.0358
99-0.0478-0.00600.0156
100-0.0511-0.0328-0.0221
101-0.0368-0.0088-0.0468
1020.01500.0435-0.0297
1030.00120.0419-0.0095
104-0.1130-0.0888-0.0646
105-0.0624-0.07790.0257
106-0.0729-0.0717-0.0382
107-0.0660-0.0312-0.0444
1080.11670.17140.2168
109-0.0067-0.00950.0401
110-0.0664-0.0674-0.0590
111-0.0726-0.0126-0.0408
112-0.0303-0.0216-0.0400
1130.08530.09870.0452
114-0.0784-0.0596-0.0572
115-0.0383-0.0030-0.0351
116-0.0129-0.00420.0243
117-0.0078-0.0587-0.0629
118-0.0376-0.04040.0024
119-0.0454-0.0404-0.0510
120-0.0184-0.0416-0.0416
1210.01770.0185-0.0269
122-0.0948-0.1037-0.0468
123-0.0142-0.02500.0867
1240.06020.04910.0327
125-0.0707-0.0723-0.0482
1260.13570.13350.1098
127-0.0195-0.0134-0.0510
1280.00390.00290.0009
1290.07620.10990.0913
130-0.0341-0.0609-0.0498
1310.04700.0143-0.0303
1320.05110.0503-0.0523
1330.0018-0.03900.0325
1340.08390.05930.0210
135-0.0202-0.0008-0.0499
136-0.0222-0.0120-0.0523
1370.0718-0.0109-0.0065
1380.02320.0312-0.0092
1390.04740.0550-0.0354
140-0.0335-0.0182-0.0584
1410.04740.1025-0.0265
142-0.0642-0.0500-0.0428
1430.03940.02270.0031
144-0.0955-0.0919-0.0516
1450.03080.02450.0881
1460.16920.18380.1093
147-0.0148-0.02080.0197
1480.10500.13660.0919
1490.07040.05620.1973
1500.12180.1118-0.0040
151-0.0532-0.04040.0200
152-0.0388-0.0156-0.0550
1530.01100.01680.0242
1540.01210.02240.0438
155-0.0230-0.0366-0.0646
1560.0328-0.0036-0.0050
1570.08860.06710.0351
158-0.02350.0087-0.0617
1590.07660.06180.0346
1600.02920.04500.1216
1610.02610.0022-0.0600
162-0.0842-0.0671-0.0575
163-0.0806-0.0581-0.0552
164-0.0485-0.0120-0.0435
165-0.0094-0.02000.0002
1660.00780.0349-0.0261
167-0.1371-0.1397-0.0628
1680.06050.06330.0880
1690.11600.0807-0.0152
170-0.0250-0.0135-0.0464
171-0.0257-0.0531-0.0436
1720.07880.09690.0356
1730.02010.07110.0969
174-0.0190-0.0167-0.0509
175-0.0653-0.0826-0.0821
1760.07400.05840.1090
177-0.0562-0.0295-0.0408
178-0.0059-0.0230-0.0440
1790.07770.08260.0853
1800.19400.18750.1345
181-0.0711-0.0394-0.0252
182-0.0613-0.0696-0.0428
183-0.0853-0.0463-0.0919
184-0.00410.0441-0.0105
1850.08720.0528-0.0031
186-0.0616-0.0481-0.0598
1870.08680.04320.0818
1880.02520.01450.0748
1890.06130.03740.0684
1900.00970.02510.0212
1910.04180.03320.0512
1920.12910.13480.0683
1930.04650.06830.0425
1940.02370.0286-0.0050
195-0.0052-0.0137-0.0238
1960.03960.08450.0920
1970.15750.12230.1163
198-0.0228-0.0174-0.0479
1990.05360.0692-0.0267
200-0.0299-0.0416-0.0573
2010.01500.0391-0.0025
202-0.0624-0.0699-0.1030
2030.07440.07170.0454
2040.06670.01490.1277
205-0.0635-0.06680.0496
2060.01920.05620.0168
207-0.03040.00310.0507
208-0.0376-0.0671-0.0514
209-0.0418-0.01580.0122
2100.16410.10740.1705
211-0.0182-0.04470.0153
2120.05910.07700.1877
213-0.0217-0.00470.0842
2140.25670.26210.2839
215-0.0266-0.0245-0.0416
216-0.0439-0.0463-0.0593
217-0.0879-0.0338-0.0464
218-0.0227-0.0882-0.0280
219-0.03460.0002-0.0591
220-0.0817-0.0689-0.0547
2210.03900.00220.0295
222-0.04030.00370.0509
223-0.0704-0.0475-0.0397
2240.00010.00920.0677
2250.05070.03340.0769
226-0.0507-0.02060.0230
2270.03430.10310.0678
228-0.0150-0.0314-0.0787
2290.00490.0038-0.0543
2300.06200.05000.0845
2310.04300.06210.0327
232-0.04100.01610.0312
233-0.0264-0.0571-0.0684
2340.18230.18510.0175
2350.20280.18010.0648
236-0.0241-0.0306-0.0425
237-0.0197-0.0328-0.0506
2380.04100.02980.0779
239-0.00520.01790.0057
2400.09150.10310.1859
241-0.00810.01440.0920
242-0.0432-0.0371-0.0491
243-0.0605-0.0730-0.0637
244-0.0470-0.0708-0.0526
245-0.0780-0.0377-0.0471
246-0.0126-0.0270-0.0401
247-0.0562-0.0807-0.0417
248-0.0642-0.0609-0.0566
249-0.0828-0.0590-0.0712
Source: HUMANENESS_SMALL_VOICE_IDENTITY_R1_R2_2026-10-02_mediathek_0047.html · SHA-256 c9330d9714011f2b…
g07GermanMediathek voice study

mediathek_0047 · de · 26/50

Generated TTS prediction · Rank 1 qkvo · 10.5 s

Dieser kleine Knoten auf meinem Schädel, das ist mehr als nur eine kleine lästige Sache. (soft hum) Glauben Sie mir, Dinge, die harmlos aussehen, können überraschend ernst sein.

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Contempt
z +2.79 / +2.81raw 1.29 / 1.30
Sourness
z +2.56 / +2.62raw 1.12 / 1.15
Disgust
z +2.13 / +2.34raw 1.05 / 1.15
Bitterness
z +2.33 / +2.16raw 0.98 / 0.91
Anger
z +2.09 / +2.14raw 1.29 / 1.32
Disappointment
z +1.74 / +1.94raw 1.22 / 1.36
Impatience and Irritability
z +1.76 / +1.94raw 1.39 / 1.52

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Authoritative style (S_AUTH)
z +2.73 / +2.59raw 3.94 / 3.79
Newsreader style (S_NEWS)
z +2.72 / +2.49raw 1.95 / 1.82
Cartoonish style (S_CART)
z +2.35 / +2.57raw 4.85 / 5.17
Tension (TENS)
z +2.35 / +2.38raw 3.56 / 3.58
Nasal resonance (R_NASL)
z +1.91 / +1.81raw 4.79 / 4.69
Metallic character (METL)
z +1.72 / +1.69raw 3.27 / 3.24
Ranting or angry style (S_RANT)
z +1.69 / +1.63raw 3.40 / 3.32

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z -0.55 / -0.65raw 1.44 / 1.28
Vocal-burst blend (0–10)z -1.39 / -1.43raw 2.86 / 2.76
Content enjoyment (CE)z +0.37 / +0.49raw 5.18 / 5.30
Production quality (PQ)z +0.19 / +0.29raw 7.21 / 7.30
Speech signal quality, calibratedz +0.03 / +0.27raw 3.50 / 3.62
Background quality, calibratedz +0.52 / +0.57raw 3.87 / 3.92
Overall quality, calibratedz +0.49 / +0.68raw 3.14 / 3.25
P.808 MOS estimatez +0.87 / +0.80raw 3.85 / 3.82
score content enjoymentz +0.72 / +0.79raw 4.72 / 4.75
score overall qualityz +0.53 / +0.86raw 2.92 / 3.09
Background Noisez -0.63 / -0.22raw 0.64 / 0.75
Tempo (TEMP)z +0.20 / +0.21raw 2.36 / 2.37
Arousal (AROU)z +1.36 / +1.29raw 3.92 / 3.84
Valence (VALN)z +0.30 / +0.20raw 2.66 / 2.55
Recording quality (RCQL)z -0.69 / -0.67raw 2.14 / 2.16
Background noise (BKGN)z -0.02 / -0.09raw 2.69 / 2.63
talking speedz +1.19 / +0.87raw 13.50 / 12.30
Characters per second (CPS): 16.84 raw · z +2.08

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8600.869
Identity · 250D0.7960.777

Vocal-burst regions

Tiny
Base
0 s10.5 s

Verified burst intervals are unavailable for this clip.

Tiny: 0 regions, count-head estimate 0.00; Base: 0 regions, count-head estimate 0.00.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.059+0.090.032+0.040.0070.570
1Amusementemo_Amusement0.426+0.480.378+0.43-0.0010.885
2Angeremo_Anger1.285+2.091.317+2.140.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.118+0.200.048+0.070.0050.572
4Aweemo_Awe0.021+0.05-0.004-0.010.0000.427
5Bitternessemo_Bitterness0.982+2.330.913+2.160.0130.415
6Concentrationemo_Concentration1.079+0.601.067+0.580.6340.741
7Confusionemo_Confusion0.182+0.340.099+0.180.0010.540
8Contemplationemo_Contemplation0.842+1.270.774+1.160.0820.598
9Contemptemo_Contempt1.294+2.791.301+2.81-0.0000.464
10Contentmentemo_Contentment-0.031-0.06-0.041-0.080.0000.514
11Disappointmentemo_Disappointment1.218+1.741.361+1.940.0000.700
12Disgustemo_Disgust1.050+2.131.153+2.340.0070.489
13Distressemo_Distress1.295+1.581.372+1.670.0000.821
14Doubtemo_Doubt0.248+0.520.223+0.470.0060.466
15Elationemo_Elation-0.050-0.09-0.056-0.090.0120.728
16Embarrassmentemo_Embarrassment-0.003-0.11-0.001-0.100.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.528+0.570.312+0.340.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.132+0.190.120+0.170.0280.539
19Fearemo_Fear0.845+1.360.809+1.300.0150.610
20Helplessnessemo_Helplessness0.771+1.110.724+1.05-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.018-0.130.036-0.110.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability1.391+1.761.519+1.940.1490.705
23Infatuationemo_Infatuation0.003+0.03-0.025-0.02-0.0150.533
24Interestemo_Interest1.713+0.141.430-0.181.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness-0.042-0.03-0.032-0.01-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.354+0.670.516+1.030.0590.442
27Longingemo_Longing0.927+1.330.982+1.420.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.845+1.480.947+1.670.0550.534
29Painemo_Pain0.281+0.490.145+0.25-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy-0.020-0.06-0.023-0.060.0170.614
31Prideemo_Pride0.288+0.560.608+1.190.0020.509
32Reliefemo_Relief0.148+0.170.130+0.140.0330.681
33Sadnessemo_Sadness1.157+1.531.125+1.480.0000.759
34Sexual Lustemo_Sexual_Lust0.022-0.020.014-0.030.0300.494
35Shameemo_Shame0.128+0.230.341+0.640.0130.510
36Sournessemo_Sourness1.120+2.561.147+2.62-0.0050.439
37Teasingemo_Teasing0.625+1.150.606+1.110.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.345+0.440.295+0.360.0410.697
39Triumphemo_Triumph0.136+0.270.320+0.63-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg4.089+0.893.932+0.772.9841.238
41Arousal (AROU)vn_AROU_reg3.915+1.363.845+1.292.3591.147
42Arousal shift (ARSH)vn_ARSH_reg3.252+0.773.180+0.692.5800.867
43Vocal attack (ATCK)vn_ATCK_reg4.341+1.684.258+1.622.1621.295
44Background noise (BKGN)vn_BKGN_reg2.686-0.022.632-0.092.7030.761
45Brightness (BRGT)vn_BRGT_reg1.960-0.041.749-0.301.9910.813
46Speech chunking (CHNK)vn_CHNK_reg2.040+0.162.030+0.151.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg3.118+1.163.283+1.342.0180.945
48Cognitive load (COGL)vn_COGL_reg2.417+0.302.363+0.252.0941.061
49Dynamic arc (DARC)vn_DARC_reg3.507+1.373.592+1.431.6151.380
50Disfluency (DFLU)vn_DFLU_reg1.862-0.531.964-0.442.4881.184
51Emphasis (EMPH)vn_EMPH_reg4.252+1.084.399+1.182.6071.524
52Esthetics (ESTH)vn_ESTH_reg1.942-0.591.753-0.832.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg-0.001-0.270.044-0.130.0860.322
54Focus (FOCS)vn_FOCS_reg2.679+0.302.753+0.372.3441.102
55Fullness (FULL)vn_FULL_reg1.123-0.741.009-0.871.7780.889
56Perceived gender (GEND)vn_GEND_reg5.026+0.615.101+0.664.0551.588
57Harmonicity (HARM)vn_HARM_reg3.128+0.773.209+0.872.4800.836
58Metallic character (METL)vn_METL_reg3.270+1.723.240+1.691.6210.958
59Pitch range (RANG)vn_RANG_reg3.759+0.943.804+0.992.9140.897
60Recording quality (RCQL)vn_RCQL_reg2.143-0.692.157-0.672.7540.888
61Vocal register (REGS)vn_REGS_reg0.206-0.780.485-0.561.1981.268
62Respiration (RESP)vn_RESP_reg2.740+0.062.780+0.092.6801.056
63Roughness (ROUG)vn_ROUG_reg3.097+1.203.077+1.182.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg3.401+1.503.202+1.262.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg1.938-0.261.994-0.192.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg1.607-0.301.636-0.271.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg0.945-0.580.817-0.741.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg4.789+1.914.690+1.812.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg2.734+0.742.658+0.642.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg3.345+1.593.100+1.282.0700.803
71Smoothness (SMTH)vn_SMTH_reg2.445+0.282.295+0.142.1431.093
72Stance (STNC)vn_STNC_reg4.309+1.474.024+1.222.6331.137
73Speech structure (STRU)vn_STRU_reg2.930+0.833.092+1.002.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg-0.115-1.05-0.018-0.981.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg3.939+2.733.792+2.591.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg4.846+2.355.167+2.571.5601.401
77Casual style (S_CASU)vn_S_CASU_reg1.428-1.851.438-1.843.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg-0.118-2.06-0.159-2.092.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg4.132+1.234.123+1.222.7211.148
80Formal style (S_FORM)vn_S_FORM_reg1.927+0.721.990+0.781.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg3.126+0.232.957+0.102.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg2.579+0.882.546+0.861.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg1.951+2.721.822+2.490.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg1.834-0.391.998-0.272.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg3.398+1.693.318+1.631.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg4.487+1.474.182+1.252.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg-0.746-1.14-0.665-1.070.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg0.930-0.411.099-0.311.5971.618
89Tempo (TEMP)vn_TEMP_reg2.355+0.202.369+0.212.1171.190
90Tension (TENS)vn_TENS_reg3.558+2.353.584+2.381.5590.852
91Valence (VALN)vn_VALN_reg2.663+0.302.552+0.202.3281.118
92Valence shift (VALS)vn_VALS_reg2.127-0.361.841-0.632.5181.079
93Velocity flux (VFLX)vn_VFLX_reg2.102-0.131.846-0.342.2561.208
94Volatility (VOLT)vn_VOLT_reg2.601+0.022.701+0.132.5860.886
95Vulnerability (VULN)vn_VULN_reg1.146-1.530.852-1.832.6480.984
96Warmth (WARM)vn_WARM_reg1.941-0.701.814-0.842.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_61.438-0.551.281-0.652.3321.621
98Vocal-burst blend (0–10)blend_0_102.864-1.392.758-1.436.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.536-1.650.495-1.781.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.233-0.063.198-0.183.2500.293
101Arousaleiv_extra:Arousal1.722+1.001.896+1.401.2890.435
102Authenticityeiv_extra:Authenticity2.705+0.252.659-0.052.6670.153
103Background Noiseeiv_extra:Background_Noise0.640-0.630.749-0.220.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.040+0.001.134+0.311.0400.303
105Gendereiv_extra:Gender-1.192-0.57-0.989-0.37-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched2.151+1.292.153+1.301.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive2.298+0.932.421+1.201.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.077-0.202.043-0.302.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous0.805-0.360.951+0.060.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.205-0.240.040-0.750.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.853+1.480.919+1.680.3760.324
112Valenceeiv_extra:Valence-0.740-1.96-1.014-2.680.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.606-0.251.439-0.631.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold-0.317-1.23-0.338-1.300.0510.300
115score background qualityeiv_extra:score_background_quality3.583+0.413.713+0.593.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.724+0.724.745+0.794.5070.303
117score overall qualityeiv_extra:score_overall_quality2.921+0.533.088+0.862.6520.507
118score speech qualityeiv_extra:score_speech_quality1.921+0.781.944+0.941.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.181+0.375.296+0.494.7991.018
120Content usefulness (CU)audiobox:CU5.768+0.125.906+0.265.6491.001
121Production complexity (PC)audiobox:PC1.473-0.531.662-0.412.3311.617
122Production quality (PQ)audiobox:PQ7.212+0.197.296+0.297.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw3.933+0.054.131+0.353.9000.669
124Background quality, rawdnsmos:BAK_raw4.021+0.724.052+0.753.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.561+0.593.723+0.793.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.504+0.033.624+0.273.4880.501
127Background quality, calibrateddnsmos:BAK3.872+0.523.919+0.573.3890.928
128Overall quality, calibrateddnsmos:OVRL3.136+0.493.252+0.682.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.855+0.873.820+0.803.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.002+0.00-0.005-0.010.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.859-0.160.730-0.390.9470.554
132Agevoiceclap_attribute:Age3.202-0.053.280+0.093.2290.547
133Amusementvoiceclap_attribute:Amusement0.148-0.430.101-0.510.3920.572
134Angervoiceclap_attribute:Anger1.929+1.851.905+1.800.9920.506
135Arousalvoiceclap_attribute:Arousal2.432+1.632.252+1.341.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.524+0.030.407-0.210.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.667-0.622.664-0.642.7700.167
138Awevoiceclap_attribute:Awe0.796+0.990.687+0.570.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.590+0.090.683+0.440.5670.264
140Bitternessvoiceclap_attribute:Bitterness1.063+2.001.010+1.770.5920.236
141Concentrationvoiceclap_attribute:Concentration1.355+0.271.441+0.431.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.027-0.090.925-0.311.0690.468
143Confusionvoiceclap_attribute:Confusion0.736+0.280.577-0.200.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.677-0.250.575-0.500.7800.406
145Contemptvoiceclap_attribute:Contempt1.327+1.831.213+1.490.7090.339
146Contentmentvoiceclap_attribute:Contentment0.321-0.870.465-0.400.5880.307
147Disappointmentvoiceclap_attribute:Disappointment1.287+1.811.289+1.820.7600.291
148Disgustvoiceclap_attribute:Disgust1.066+2.131.029+1.950.6250.207
149Distressvoiceclap_attribute:Distress1.446+1.481.473+1.560.9270.350
150Doubtvoiceclap_attribute:Doubt0.398-0.730.352-0.970.5350.188
151Elationvoiceclap_attribute:Elation0.517-0.060.565+0.060.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.141-0.530.082-0.720.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.561-0.390.698+0.130.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.418-0.170.540+0.300.4620.261
155Fearvoiceclap_attribute:Fear0.857+1.130.925+1.450.6180.212
156Gendervoiceclap_attribute:Gender-1.658-1.50-1.226-1.07-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.944+1.060.957+1.110.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched2.418+1.542.301+1.231.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.917+0.530.980+0.700.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability2.323+2.352.221+2.161.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.570-0.110.645+0.110.6080.332
162Interestvoiceclap_attribute:Interest1.496-0.231.482-0.261.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.549+0.100.572+0.150.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.904+1.560.932+1.670.5230.245
165Longingvoiceclap_attribute:Longing0.708+0.770.719+0.830.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice1.571+2.131.533+2.020.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive2.159+0.652.120+0.581.7750.594
168Painvoiceclap_attribute:Pain0.726+1.580.763+1.750.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.393-0.370.382-0.400.5030.301
170Pridevoiceclap_attribute:Pride1.185+1.161.207+1.220.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality1.988-0.992.073-0.732.3130.328
172Reliefvoiceclap_attribute:Relief0.892-0.320.875-0.361.0250.415
173Sadnessvoiceclap_attribute:Sadness1.052+0.961.058+0.980.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.376-0.750.269-0.940.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.545-0.210.588-0.110.6350.419
176Shamevoiceclap_attribute:Shame0.607+0.280.745+0.770.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh-0.282-1.23-0.375-1.480.1600.360
178Sournessvoiceclap_attribute:Sourness1.225+1.791.168+1.590.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant1.206+1.631.285+1.840.6230.359
180Teasingvoiceclap_attribute:Teasing0.232-0.810.226-0.830.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.811+0.812.082+1.321.3840.529
182Triumphvoiceclap_attribute:Triumph1.118+1.121.239+1.450.7160.359
183Valencevoiceclap_attribute:Valence-1.790-2.81-1.701-2.67-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached1.715+0.141.776+0.251.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold-0.618-1.38-0.725-1.65-0.0660.400
186durationvoiceclap_attribute:duration10.952+0.0111.619+0.1110.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.445+0.403.620+0.753.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.669+0.094.678+0.154.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.919+0.693.029+0.952.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.961+1.011.941+0.721.8910.069
191talking speedvoiceclap_attribute:talking_speed13.498+1.1912.296+0.879.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.04980.00710.0383
1-0.00020.0171-0.0816
2-0.0720-0.0464-0.0806
3-0.0343-0.1012-0.0833
4-0.0620-0.1214-0.0772
5-0.1469-0.1326-0.1039
6-0.0327-0.0408-0.0745
70.07630.05090.1173
80.0297-0.02390.1211
9-0.1201-0.0923-0.0672
100.05550.02810.0937
11-0.0544-0.0590-0.0406
12-0.0561-0.0271-0.0191
13-0.0894-0.0856-0.0811
14-0.0769-0.0766-0.0799
15-0.1110-0.1029-0.0736
16-0.0831-0.0727-0.0872
17-0.0911-0.0996-0.0413
18-0.0272-0.0092-0.0629
19-0.0677-0.0565-0.0777
200.16850.15210.0917
21-0.0236-0.0139-0.0155
22-0.0911-0.0805-0.0772
23-0.0292-0.07220.0719
24-0.0854-0.0710-0.0663
25-0.0379-0.0621-0.0518
26-0.0223-0.0082-0.0656
27-0.0364-0.00040.1190
28-0.0265-0.01300.0341
290.00580.11800.1088
300.15940.13980.1671
310.0330-0.0162-0.0688
32-0.0677-0.0794-0.0791
33-0.1435-0.1063-0.0905
34-0.0797-0.0593-0.0736
350.21060.27300.2346
36-0.0865-0.0940-0.0883
37-0.1104-0.1175-0.0972
38-0.0527-0.0621-0.0519
39-0.1173-0.0889-0.0828
40-0.0888-0.0386-0.0151
41-0.1275-0.0934-0.1180
420.12980.10680.0573
430.05270.05680.0630
440.24250.29530.1798
450.09850.13030.1481
46-0.0905-0.0668-0.0635
47-0.0786-0.0817-0.0830
48-0.0370-0.0512-0.0749
490.08430.10180.0543
500.06720.04730.0645
51-0.0905-0.0945-0.0821
520.00400.01920.0083
530.08310.14260.0891
54-0.1287-0.1063-0.0919
550.16960.16890.1279
560.0356-0.0150-0.0762
57-0.1315-0.1555-0.0901
580.06600.01490.0702
590.09220.05010.0775
60-0.0769-0.0722-0.0792
610.26870.20140.2468
62-0.0996-0.0699-0.0768
63-0.0310-0.0828-0.0780
640.00170.02490.0606
65-0.0996-0.0856-0.0670
660.07340.09340.0624
67-0.0865-0.0850-0.0792
68-0.1053-0.1096-0.0868
69-0.0820-0.0811-0.0753
700.07400.09840.1000
71-0.0106-0.0248-0.0171
72-0.0734-0.0439-0.0795
730.10130.06460.1074
74-0.1361-0.0557-0.0897
750.04670.04140.1395
76-0.0272-0.00050.0095
77-0.0272-0.0266-0.0750
780.0584-0.0200-0.0185
79-0.0808-0.0744-0.0712
80-0.0292-0.0200-0.0585
810.05720.06940.0006
82-0.0882-0.1158-0.0786
83-0.0951-0.1018-0.0841
840.09560.16000.1237
850.03790.07770.0580
86-0.0985-0.1091-0.0751
87-0.0586-0.0590-0.0576
88-0.0048-0.0450-0.0311
89-0.0905-0.0805-0.0754
90-0.1030-0.1113-0.0891
910.10930.08170.1553
92-0.1087-0.1001-0.0798
930.14800.16440.1699
940.14290.08050.0975
950.02950.0545-0.0818
96-0.0601-0.0540-0.0734
970.00310.02140.0243
980.10420.09510.1093
990.0444-0.00430.0981
1000.12920.12700.0787
1010.07290.07660.0853
102-0.0178-0.0281-0.0364
103-0.1076-0.0951-0.0867
104-0.0376-0.0192-0.0803
105-0.0319-0.0375-0.0357
106-0.0549-0.0475-0.0959
1070.00950.0298-0.0210
1080.02860.0142-0.0374
109-0.0549-0.0940-0.0806
1100.0324-0.0179-0.0271
111-0.1053-0.0750-0.0953
112-0.0854-0.0833-0.0809
1130.04180.08280.0481
114-0.0917-0.0554-0.0030
1150.04330.09170.1879
116-0.0085-0.0011-0.0704
1170.07510.07550.1299
118-0.1053-0.1124-0.0800
1190.05840.12860.0688
120-0.1036-0.0917-0.0845
1210.10760.10350.1042
122-0.0566-0.0403-0.0844
123-0.1116-0.0543-0.0695
1240.0515-0.0022-0.0596
1250.02080.07100.0840
126-0.0467-0.0559-0.0033
1270.01060.01620.0614
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.0317-0.0287-0.0478
1-0.0453-0.0265-0.0421
20.03010.01590.0653
3-0.00900.00660.0638
40.02150.0422-0.0395
5-0.0683-0.0369-0.0454
60.00170.00760.0047
70.01110.0374-0.0440
80.01850.0262-0.0207
90.03380.01580.0231
100.05160.05800.0140
11-0.0531-0.0577-0.0234
12-0.0534-0.0619-0.0475
13-0.0631-0.0237-0.0384
14-0.0559-0.0527-0.0112
150.06220.05400.1014
160.06770.06060.0463
17-0.0370-0.0482-0.0340
18-0.0447-0.0348-0.0442
190.0037-0.0179-0.0423
200.09120.07850.0520
21-0.0620-0.0651-0.0403
22-0.0280-0.0461-0.0451
23-0.0998-0.0654-0.0420
240.03670.06800.0601
25-0.01780.0216-0.0799
260.07030.02120.0520
270.09060.04560.0311
280.04820.05350.0812
29-0.0792-0.0775-0.0418
300.02610.0519-0.0187
310.04680.07120.0295
320.06830.0859-0.0404
330.05650.01550.0334
34-0.0513-0.0238-0.0482
35-0.0181-0.0369-0.0560
360.02800.00940.0763
370.03310.03980.0815
380.02950.07750.0651
39-0.0562-0.0324-0.0317
40-0.0740-0.0590-0.0503
410.12850.09750.0320
42-0.0597-0.0237-0.0432
430.07080.1181-0.0219
440.07690.0661-0.0079
45-0.0502-0.0351-0.0142
46-0.0597-0.0656-0.0418
470.0238-0.0242-0.0174
480.02600.0123-0.0464
49-0.0361-0.0453-0.0693
50-0.0238-0.0336-0.0786
510.07060.05400.0656
52-0.00930.0312-0.0356
530.03300.05430.0541
540.04040.05770.0339
550.05620.04110.1316
560.04990.08850.1090
570.0250-0.0235-0.0113
580.01300.0119-0.0384
59-0.0153-0.01450.0470
600.06450.08330.0616
61-0.0485-0.0461-0.0403
62-0.0516-0.0408-0.0382
630.0318-0.02190.0282
64-0.0191-0.0215-0.0362
650.0222-0.0206-0.0058
660.02710.0121-0.0473
67-0.0121-0.01770.0775
68-0.0568-0.0387-0.0468
69-0.01340.08750.0262
70-0.0611-0.0672-0.0630
71-0.0490-0.0300-0.0421
720.02410.02320.0095
730.10440.06220.0975
74-0.0568-0.0310-0.0469
750.03240.07540.0534
760.05050.09330.0421
77-0.0473-0.0622-0.0443
78-0.0228-0.0503-0.0627
790.01850.02620.0665
80-0.0786-0.0875-0.0461
810.01880.0453-0.0152
82-0.0148-0.0527-0.0295
83-0.0599-0.0325-0.0386
840.04590.04480.0440
850.07140.07960.1219
860.02050.0047-0.0414
87-0.0410-0.05060.0122
880.11650.08430.1407
890.01030.04510.0146
90-0.0222-0.03850.0137
91-0.0680-0.0743-0.0546
92-0.0132-0.01980.0359
930.01150.0406-0.0454
94-0.0161-0.0211-0.0475
95-0.0611-0.0073-0.0484
96-0.01310.03580.0255
97-0.0513-0.0488-0.0547
98-0.00700.03450.0098
99-0.00770.02850.0202
100-0.0447-0.04270.0343
101-0.00230.0160-0.0430
1020.04100.05980.0354
103-0.01200.0152-0.0544
104-0.0496-0.0632-0.0605
105-0.0519-0.0585-0.0388
106-0.0815-0.0564-0.0357
107-0.0706-0.0453-0.0416
1080.19730.21290.2717
1090.00290.0052-0.0138
110-0.0413-0.0393-0.0552
111-0.0752-0.0269-0.0382
112-0.0399-0.0366-0.0374
1130.05940.06250.0984
114-0.0769-0.0543-0.0647
115-0.00590.00720.0304
1160.0018-0.0282-0.0051
117-0.0078-0.02150.0574
118-0.0465-0.0459-0.0453
119-0.0565-0.0419-0.0477
1200.0005-0.0096-0.0176
121-0.0084-0.00360.0092
122-0.0998-0.1086-0.0438
1230.00890.01290.0374
1240.13080.06960.0042
125-0.0992-0.1117-0.0451
1260.09240.09750.1187
127-0.0294-0.0358-0.0477
1280.01730.02670.0277
1290.09470.09330.0917
130-0.0354-0.0073-0.0466
1310.0244-0.0055-0.0305
1320.06340.0630-0.0096
1330.0147-0.02950.0378
1340.04130.02740.0496
135-0.02220.0034-0.0421
1360.0020-0.01990.0196
1370.0599-0.00800.0341
1380.05110.0273-0.0409
1390.02410.04800.0370
140-0.02270.0055-0.0412
1410.08320.11230.0371
142-0.0459-0.0157-0.0401
1430.05880.01170.0148
144-0.0929-0.0864-0.0483
1450.03960.01920.0301
1460.13480.17820.0879
147-0.0328-0.0252-0.0121
1480.14570.16020.1742
1490.13940.12700.1294
1500.10380.10120.0920
1510.00740.00370.0456
1520.01160.01710.0159
153-0.00730.02580.0338
1540.06770.04900.0985
1550.0131-0.04820.0006
1560.0151-0.0128-0.0387
1570.02190.04590.0249
1580.04790.05960.0577
159-0.00030.0115-0.0508
1600.05770.05240.1270
161-0.0217-0.0145-0.0610
162-0.1090-0.0817-0.0538
163-0.0640-0.0556-0.0516
164-0.01740.0090-0.0407
165-0.0536-0.0076-0.0118
1660.04620.09700.0298
167-0.1400-0.1803-0.0588
168-0.00830.02950.0829
1690.08830.04740.0219
170-0.0054-0.01900.0156
171-0.0202-0.0343-0.0408
1720.07110.08640.0555
1730.08610.11070.0707
174-0.0482-0.0472-0.0476
175-0.0508-0.0493-0.0694
1760.08890.02530.0812
177-0.0513-0.0372-0.0382
1780.03030.0042-0.0204
1790.09750.15070.1061
1800.20880.17080.1488
181-0.0668-0.03580.0191
182-0.0470-0.0480-0.0400
183-0.0651-0.0596-0.0680
184-0.02800.03660.0135
1850.05050.01960.0726
186-0.0479-0.0340-0.0559
1870.07630.07270.0708
1880.05480.04080.0748
1890.05790.01450.0713
1900.03990.02600.0800
191-0.0205-0.00480.0857
1920.05160.09170.1123
1930.02910.06320.1193
1940.04820.0459-0.0066
195-0.0142-0.0139-0.0441
1960.06400.10170.0850
1970.14110.12760.0864
198-0.0202-0.0169-0.0341
1990.05160.07430.0350
200-0.0591-0.0614-0.0536
2010.06050.07220.0598
202-0.0479-0.0459-0.0965
2030.13940.10010.0313
2040.06860.04800.1166
205-0.0425-0.0329-0.0047
2060.02090.04060.0328
207-0.0307-0.0203-0.0383
208-0.0166-0.0177-0.0299
209-0.0447-0.01990.0224
2100.21230.17180.2505
2110.0229-0.0256-0.0356
2120.04530.06040.0547
2130.06540.04220.0781
2140.27310.28250.2739
215-0.0272-0.0366-0.0390
2160.0140-0.00620.0281
217-0.0694-0.0472-0.0516
218-0.0482-0.05060.0255
219-0.0511-0.0427-0.0604
220-0.0511-0.0482-0.0512
2210.01410.00830.0170
222-0.00730.0051-0.0234
223-0.0479-0.0343-0.0372
224-0.0321-0.0190-0.0243
2250.09180.07270.1033
226-0.0308-0.0115-0.0296
2270.02050.0511-0.0394
2280.00070.0034-0.0751
2290.03960.05190.0057
2300.01170.00780.0189
2310.06400.04320.0692
232-0.0222-0.0165-0.0105
233-0.0010-0.0335-0.0484
2340.21570.22350.1165
2350.13370.19080.1105
236-0.0146-0.0178-0.0097
2370.03010.0171-0.0250
2380.06770.02520.0748
239-0.0240-0.0440-0.0479
2400.02830.06590.0720
241-0.00230.01910.0504
2420.01340.02640.0159
243-0.0490-0.0680-0.0591
244-0.0376-0.0493-0.0493
245-0.0826-0.0335-0.0441
246-0.0393-0.0448-0.0375
247-0.0686-0.0891-0.0390
248-0.0631-0.0585-0.0530
249-0.0387-0.0337-0.0599
Source: HUMANENESS_SMALL_VOICE_IDENTITY_R1_R2_2026-10-02_mediathek_0047.html · SHA-256 bab76f5811287dbb…
g08GermanMediathek voice study

mediathek_0047 · de · 44/50

Generated TTS prediction · Rank 2 qv · 8.4 s

Ich fühle einfach... (slow breathing) nichts mehr, wenn ich daran denke. Jeder Arbeitgeber muss sicherstellen, dass die Dinge anders sind.

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Fatigue Exhaustion
z +4.36 / +4.36raw 2.38 / 2.38
Disappointment
z +2.42 / +2.53raw 1.70 / 1.77
Sadness
z +2.23 / +2.47raw 1.69 / 1.87
Contemplation
z +2.46 / +2.05raw 1.55 / 1.31
Helplessness
z +2.05 / +2.21raw 1.42 / 1.54
Doubt
z +1.90 / +2.04raw 0.89 / 0.96
Longing
z +1.94 / +1.70raw 1.31 / 1.16

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Cognitive load (COGL)
z +1.73 / +1.62raw 3.93 / 3.82
Chest resonance (R_CHST)
z +1.64 / +1.72raw 3.52 / 3.59
Tension (TENS)
z +1.39 / +1.32raw 2.74 / 2.68
Perceived voice age (AGEV)
z +1.32 / +1.23raw 4.62 / 4.51
Disfluency (DFLU)
z +1.00 / +1.12raw 3.67 / 3.81
Whispered talking style (S_WHIS)
z +0.87 / +1.11raw 3.01 / 3.39
Perceived gender (GEND)
z +0.83 / +0.96raw 5.37 / 5.59

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z -0.05 / +0.01raw 2.26 / 2.35
Vocal-burst blend (0–10)z -0.58 / -1.13raw 5.09 / 3.59
Content enjoyment (CE)z +0.08 / +0.16raw 4.88 / 4.96
Production quality (PQ)z +0.46 / +0.59raw 7.45 / 7.57
Speech signal quality, calibratedz +0.05 / +0.05raw 3.51 / 3.51
Background quality, calibratedz +0.80 / +0.41raw 4.13 / 3.77
Overall quality, calibratedz +0.61 / +0.39raw 3.21 / 3.08
P.808 MOS estimatez +0.57 / +0.12raw 3.70 / 3.48
score content enjoymentz -0.25 / +0.15raw 4.43 / 4.55
score overall qualityz +1.36 / +0.76raw 3.34 / 3.04
Background Noisez -1.79 / -1.04raw 0.33 / 0.53
Tempo (TEMP)z -0.95 / -0.80raw 0.98 / 1.17
Arousal (AROU)z -1.25 / -1.18raw 0.92 / 1.00
Valence (VALN)z -0.72 / -0.32raw 1.53 / 1.97
Recording quality (RCQL)z -0.53 / -0.56raw 2.28 / 2.25
Background noise (BKGN)z -0.26 / -0.28raw 2.51 / 2.49
talking speedz +0.47 / +0.12raw 10.81 / 9.49
Characters per second (CPS): 16.52 raw · z +2.00

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8960.914
Identity · 250D0.8120.794

Vocal-burst regions

Tiny
Base
0 s8.4 s

Verified burst intervals are unavailable for this clip.

Tiny: 1 regions, count-head estimate 1.08; Base: 1 regions, count-head estimate 0.96.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
Tiny12.943.120.18
Base12.943.080.14
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.093+0.150.026+0.030.0070.570
1Amusementemo_Amusement0.049+0.060.078+0.09-0.0010.885
2Angeremo_Anger0.375+0.610.523+0.850.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.004-0.00-0.009-0.030.0050.572
4Aweemo_Awe0.026+0.060.015+0.030.0000.427
5Bitternessemo_Bitterness0.792+1.870.744+1.760.0130.415
6Concentrationemo_Concentration0.504-0.180.637+0.000.6340.741
7Confusionemo_Confusion0.524+0.970.807+1.490.0010.540
8Contemplationemo_Contemplation1.555+2.461.306+2.050.0820.598
9Contemptemo_Contempt0.013+0.030.015+0.03-0.0000.464
10Contentmentemo_Contentment0.039+0.080.043+0.080.0000.514
11Disappointmentemo_Disappointment1.695+2.421.772+2.530.0000.700
12Disgustemo_Disgust0.029+0.040.032+0.050.0070.489
13Distressemo_Distress1.519+1.851.576+1.920.0000.821
14Doubtemo_Doubt0.892+1.900.956+2.040.0060.466
15Elationemo_Elation0.138+0.170.048+0.050.0120.728
16Embarrassmentemo_Embarrassment0.482+1.020.477+1.010.0430.430
17Emotional Numbnessemo_Emotional_Numbness1.574+1.711.468+1.600.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion2.378+4.362.378+4.360.0280.539
19Fearemo_Fear-0.010-0.040.100+0.140.0150.610
20Helplessnessemo_Helplessness1.421+2.051.537+2.21-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism-0.033-0.20-0.006-0.160.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.818+0.950.923+1.100.1490.705
23Infatuationemo_Infatuation-0.000+0.03-0.001+0.03-0.0150.533
24Interestemo_Interest0.257-1.470.414-1.301.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness0.008+0.060.050+0.13-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.117+0.130.117+0.130.0590.442
27Longingemo_Longing1.309+1.941.160+1.700.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.083+0.050.102+0.090.0550.534
29Painemo_Pain0.221+0.380.400+0.69-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy-0.037-0.09-0.015-0.050.0170.614
31Prideemo_Pride0.056+0.110.031+0.060.0020.509
32Reliefemo_Relief0.733+1.030.861+1.220.0330.681
33Sadnessemo_Sadness1.695+2.231.874+2.470.0000.759
34Sexual Lustemo_Sexual_Lust0.010-0.040.007-0.050.0300.494
35Shameemo_Shame0.952+1.840.854+1.650.0130.510
36Sournessemo_Sourness0.161+0.380.087+0.21-0.0050.439
37Teasingemo_Teasing-0.001-0.000.003+0.010.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.158+0.170.233+0.280.0410.697
39Triumphemo_Triumph0.016+0.040.026+0.06-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg4.621+1.324.512+1.232.9841.238
41Arousal (AROU)vn_AROU_reg0.923-1.251.002-1.182.3591.147
42Arousal shift (ARSH)vn_ARSH_reg1.136-1.661.225-1.562.5800.867
43Vocal attack (ATCK)vn_ATCK_reg0.967-0.921.239-0.712.1621.295
44Background noise (BKGN)vn_BKGN_reg2.507-0.262.488-0.282.7030.761
45Brightness (BRGT)vn_BRGT_reg1.820-0.211.932-0.071.9910.813
46Speech chunking (CHNK)vn_CHNK_reg0.450-1.130.674-0.951.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg1.312-0.751.418-0.642.0180.945
48Cognitive load (COGL)vn_COGL_reg3.932+1.733.817+1.622.0941.061
49Dynamic arc (DARC)vn_DARC_reg-0.058-1.210.314-0.941.6151.380
50Disfluency (DFLU)vn_DFLU_reg3.673+1.003.812+1.122.4881.184
51Emphasis (EMPH)vn_EMPH_reg1.906-0.462.255-0.232.6071.524
52Esthetics (ESTH)vn_ESTH_reg1.768-0.812.069-0.432.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg-0.429-1.60-0.280-1.140.0860.322
54Focus (FOCS)vn_FOCS_reg0.230-1.920.672-1.522.3441.102
55Fullness (FULL)vn_FULL_reg1.989+0.242.020+0.271.7780.889
56Perceived gender (GEND)vn_GEND_reg5.374+0.835.586+0.964.0551.588
57Harmonicity (HARM)vn_HARM_reg2.032-0.542.135-0.412.4800.836
58Metallic character (METL)vn_METL_reg0.786-0.871.087-0.561.6210.958
59Pitch range (RANG)vn_RANG_reg2.453-0.512.512-0.452.9140.897
60Recording quality (RCQL)vn_RCQL_reg2.280-0.532.252-0.562.7540.888
61Vocal register (REGS)vn_REGS_reg-0.069-1.000.039-0.911.1981.268
62Respiration (RESP)vn_RESP_reg3.548+0.823.352+0.642.6801.056
63Roughness (ROUG)vn_ROUG_reg2.539+0.572.615+0.662.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg3.520+1.643.587+1.722.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg0.915-1.501.387-0.932.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg1.123-0.781.335-0.571.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg1.660+0.271.860+0.511.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg2.049-0.712.493-0.282.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg1.950-0.322.181-0.002.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg1.975-0.122.116+0.062.0700.803
71Smoothness (SMTH)vn_SMTH_reg1.364-0.711.650-0.452.1431.093
72Stance (STNC)vn_STNC_reg1.938-0.612.238-0.352.6331.137
73Speech structure (STRU)vn_STRU_reg1.955-0.192.066-0.072.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg2.012+0.462.010+0.451.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg0.668-0.400.934-0.151.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg1.281-0.201.203-0.251.5601.401
77Casual style (S_CASU)vn_S_CASU_reg3.344-0.353.507-0.223.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg1.692-0.682.092-0.382.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg2.983+0.232.819+0.092.7211.148
80Formal style (S_FORM)vn_S_FORM_reg0.722-0.460.846-0.341.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg4.058+0.964.048+0.952.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg1.486+0.121.480+0.121.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.563+0.270.607+0.350.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg0.877-1.121.162-0.902.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg2.087+0.682.202+0.771.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg1.778-0.511.993-0.362.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg-1.527-1.77-0.784-1.170.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg3.010+0.873.391+1.111.5971.618
89Tempo (TEMP)vn_TEMP_reg0.985-0.951.168-0.802.1171.190
90Tension (TENS)vn_TENS_reg2.739+1.392.680+1.321.5590.852
91Valence (VALN)vn_VALN_reg1.528-0.721.971-0.322.3281.118
92Valence shift (VALS)vn_VALS_reg1.395-1.041.803-0.662.5181.079
93Velocity flux (VFLX)vn_VFLX_reg-0.235-2.060.086-1.802.2561.208
94Volatility (VOLT)vn_VOLT_reg2.625+0.042.650+0.072.5860.886
95Vulnerability (VULN)vn_VULN_reg3.514+0.883.075+0.432.6480.984
96Warmth (WARM)vn_WARM_reg3.287+0.783.326+0.822.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_62.257-0.052.345+0.012.3321.621
98Vocal-burst blend (0–10)blend_0_105.094-0.583.586-1.136.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.884-0.620.756-1.001.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.623+1.273.322+0.243.2500.293
101Arousaleiv_extra:Arousal0.861-0.981.138-0.351.2890.435
102Authenticityeiv_extra:Authenticity2.771+0.682.921+1.662.6670.153
103Background Noiseeiv_extra:Background_Noise0.334-1.790.534-1.040.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.503+1.531.520+1.591.0400.303
105Gendereiv_extra:Gender-0.504+0.12-0.705-0.08-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.724-0.411.842+0.061.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive1.128-1.561.674-0.391.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.140-0.012.112-0.102.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous0.688-0.710.710-0.640.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.338+0.180.476+0.610.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.186-0.590.451+0.230.3760.324
112Valenceeiv_extra:Valence-0.582-1.54-0.978-2.590.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.463-0.581.293-0.971.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold-0.003-0.18-0.041-0.310.0510.300
115score background qualityeiv_extra:score_background_quality3.968+0.963.651+0.503.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.431-0.254.553+0.154.5070.303
117score overall qualityeiv_extra:score_overall_quality3.341+1.363.036+0.762.6520.507
118score speech qualityeiv_extra:score_speech_quality1.887+0.541.913+0.721.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE4.880+0.084.960+0.164.7991.018
120Content usefulness (CU)audiobox:CU6.116+0.476.005+0.365.6491.001
121Production complexity (PC)audiobox:PC1.833-0.311.959-0.232.3311.617
122Production quality (PQ)audiobox:PQ7.445+0.467.565+0.597.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw3.947+0.073.934+0.053.9000.669
124Background quality, rawdnsmos:BAK_raw4.228+0.903.876+0.593.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.615+0.653.465+0.473.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.513+0.053.513+0.053.4880.501
127Background quality, calibrateddnsmos:BAK4.133+0.803.767+0.413.3890.928
128Overall quality, calibrateddnsmos:OVRL3.212+0.613.077+0.392.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.702+0.573.482+0.123.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.732+1.130.674+1.040.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.839-0.200.744-0.370.9470.554
132Agevoiceclap_attribute:Age3.396+0.313.362+0.243.2290.547
133Amusementvoiceclap_attribute:Amusement0.350-0.070.213-0.310.3920.572
134Angervoiceclap_attribute:Anger0.961-0.061.411+0.830.9920.506
135Arousalvoiceclap_attribute:Arousal1.270-0.241.331-0.141.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.363-0.300.389-0.250.5090.479
137Authenticityvoiceclap_attribute:Authenticity3.035+1.592.807+0.222.7700.167
138Awevoiceclap_attribute:Awe0.152-1.490.134-1.560.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.208-1.360.547-0.080.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.661+0.290.749+0.660.5920.236
141Concentrationvoiceclap_attribute:Concentration0.800-0.781.088-0.241.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.245+0.381.579+1.091.0690.468
143Confusionvoiceclap_attribute:Confusion1.012+1.101.197+1.660.6430.334
144Contemplationvoiceclap_attribute:Contemplation1.301+1.281.147+0.900.7800.406
145Contemptvoiceclap_attribute:Contempt0.539-0.500.629-0.230.7090.339
146Contentmentvoiceclap_attribute:Contentment0.206-1.250.093-1.610.5880.307
147Disappointmentvoiceclap_attribute:Disappointment1.298+1.851.323+1.940.7600.291
148Disgustvoiceclap_attribute:Disgust0.590-0.170.669+0.220.6250.207
149Distressvoiceclap_attribute:Distress0.965+0.111.002+0.220.9270.350
150Doubtvoiceclap_attribute:Doubt1.039+2.681.006+2.500.5350.188
151Elationvoiceclap_attribute:Elation0.115-1.07-0.047-1.480.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.650+1.090.539+0.740.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.936+1.021.051+1.450.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion1.228+2.941.106+2.470.4620.261
155Fearvoiceclap_attribute:Fear0.772+0.730.728+0.520.6180.212
156Gendervoiceclap_attribute:Gender-0.359-0.21-0.488-0.34-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness1.034+1.401.034+1.400.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.962+0.322.003+0.441.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.283-1.060.153-1.390.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability1.126+0.141.352+0.551.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.545-0.190.420-0.570.6080.332
162Interestvoiceclap_attribute:Interest0.555-2.370.833-1.741.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.883+0.820.809+0.660.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.532+0.040.589+0.270.5230.245
165Longingvoiceclap_attribute:Longing0.929+1.780.812+1.250.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.580-0.730.835+0.000.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive1.430-0.581.592-0.311.7750.594
168Painvoiceclap_attribute:Pain0.628+1.130.648+1.220.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.208-0.980.161-1.140.5030.301
170Pridevoiceclap_attribute:Pride0.799+0.050.646-0.390.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.164-0.461.774-1.652.3130.328
172Reliefvoiceclap_attribute:Relief1.144+0.291.268+0.591.0250.415
173Sadnessvoiceclap_attribute:Sadness1.305+1.771.198+1.420.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.616-0.330.587-0.380.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.026-1.450.066-1.360.6350.419
176Shamevoiceclap_attribute:Shame0.868+1.210.912+1.360.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.193+0.090.035-0.350.1600.360
178Sournessvoiceclap_attribute:Sourness0.710+0.020.694-0.040.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.376-0.690.476-0.410.6230.359
180Teasingvoiceclap_attribute:Teasing0.235-0.810.453-0.230.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.314-0.131.148-0.451.3840.529
182Triumphvoiceclap_attribute:Triumph0.625-0.250.511-0.570.7160.359
183Valencevoiceclap_attribute:Valence-0.740-1.16-1.235-1.94-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached1.138-0.901.349-0.521.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold-0.133-0.17-0.497-1.08-0.0660.400
186durationvoiceclap_attribute:duration10.349-0.0911.877+0.1510.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.585+0.683.608+0.733.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.652-0.024.589-0.464.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.942+0.742.966+0.802.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.932+0.601.946+0.801.8910.069
191talking speedvoiceclap_attribute:talking_speed10.808+0.479.490+0.129.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.05540.06020.0685
1-0.0376-0.0186-0.0792
2-0.0193-0.0328-0.0782
30.0110-0.0028-0.0644
4-0.0889-0.0582-0.0750
5-0.0625-0.0796-0.1009
6-0.0344-0.0532-0.0723
70.10130.11360.1131
80.02580.02260.1183
9-0.0557-0.0578-0.0435
100.04120.02480.0884
11-0.0394-0.0844-0.0677
120.00890.01240.0082
13-0.0812-0.0896-0.0787
14-0.1256-0.0968-0.0830
15-0.0604-0.0848-0.0714
16-0.0770-0.0777-0.0987
17-0.0658-0.0525-0.0449
18-0.0578-0.0405-0.0735
19-0.0634-0.0175-0.0754
200.22760.12560.0949
210.03630.06900.0090
22-0.0696-0.0757-0.0750
23-0.0316-0.06380.0222
24-0.0566-0.0762-0.0643
25-0.0433-0.0270-0.0565
26-0.0628-0.0145-0.0676
270.06840.11070.1350
280.02640.00500.0101
290.00700.04990.0877
300.21570.17260.1571
310.04240.0477-0.0235
32-0.0566-0.0978-0.0768
33-0.1156-0.1035-0.0879
34-0.0711-0.0424-0.0671
350.28920.24450.2153
36-0.0764-0.0829-0.0857
37-0.0519-0.0892-0.0944
38-0.1031-0.0786-0.0514
39-0.0865-0.1011-0.0804
40-0.0190-0.04840.0335
41-0.1031-0.1304-0.1146
420.00590.04310.0275
430.04950.06710.0638
440.19910.25310.1791
450.06900.11940.1401
46-0.0978-0.0978-0.0751
47-0.0658-0.0954-0.0806
48-0.0265-0.0040-0.0707
490.03210.01080.0208
500.11320.08580.0734
51-0.0972-0.0978-0.0797
520.0024-0.0391-0.0112
530.03070.03810.0670
54-0.1233-0.1131-0.0892
550.22520.20330.1417
56-0.0161-0.0270-0.0756
57-0.1221-0.1381-0.0874
580.0096-0.01460.0660
590.15410.11310.1045
60-0.0865-0.0872-0.0769
610.26430.20610.2571
62-0.0806-0.0844-0.0745
63-0.0361-0.0400-0.0633
640.02620.00090.0654
65-0.0824-0.0714-0.0650
66-0.00830.07290.0830
67-0.0800-0.0935-0.0769
68-0.0895-0.1112-0.0843
69-0.0726-0.1045-0.0731
700.16830.18120.1598
71-0.0304-0.0753-0.0436
72-0.0375-0.0733-0.0771
730.03080.04290.0967
74-0.1013-0.0767-0.0871
750.02390.04600.1240
76-0.00360.0134-0.0200
77-0.0378-0.0328-0.0728
780.06580.02890.0031
79-0.0652-0.0767-0.0692
80-0.0241-0.0291-0.0702
810.0403-0.0027-0.0515
82-0.0495-0.0791-0.0763
83-0.1090-0.1059-0.0816
840.09420.10690.1655
850.05870.08390.0839
86-0.0441-0.0709-0.0781
87-0.0367-0.0338-0.0396
88-0.0447-0.0174-0.0223
89-0.0483-0.0652-0.0732
90-0.0877-0.0762-0.0864
910.15760.17450.1691
92-0.1037-0.0676-0.0774
930.13220.19850.1757
940.13100.08770.1140
95-0.0602-0.0695-0.0856
96-0.0960-0.0887-0.0910
97-0.0170-0.0467-0.0299
980.09190.14380.1167
990.04590.03120.0774
1000.05810.09640.0555
1010.09960.08100.0689
102-0.0133-0.02950.0135
103-0.0616-0.0705-0.0842
104-0.0613-0.0762-0.0779
105-0.0191-0.0573-0.0595
106-0.0560-0.0549-0.0535
107-0.0462-0.0165-0.0515
108-0.0107-0.0070-0.0512
109-0.0770-0.0666-0.0782
110-0.0519-0.0590-0.0479
111-0.1002-0.1021-0.0925
112-0.0613-0.0815-0.0786
1130.08770.11650.0477
114-0.0415-0.0532-0.0325
1150.05840.04750.1431
116-0.0548-0.0369-0.0684
1170.15760.09780.1548
118-0.1730-0.1170-0.0776
1190.09420.14960.0771
120-0.0794-0.0657-0.0820
1210.00910.00240.0757
122-0.1073-0.0968-0.0819
123-0.0853-0.0563-0.0839
124-0.0307-0.0443-0.0773
1250.07110.11410.1079
126-0.0163-0.0410-0.0136
127-0.0524-0.0511-0.0067
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.0218-0.0268-0.0457
1-0.0499-0.0458-0.0402
20.03080.05410.0720
3-0.0066-0.00450.0304
4-0.01210.0297-0.0378
5-0.0664-0.0191-0.0434
6-0.00320.00540.0139
70.0049-0.0278-0.0166
80.00190.0001-0.0346
90.05270.03590.0511
100.08350.06430.0187
11-0.0608-0.0791-0.0299
12-0.0997-0.0900-0.0455
13-0.0625-0.0671-0.0367
14-0.0501-0.05050.0058
150.08960.04080.1163
160.09130.07960.0477
17-0.0437-0.0606-0.0499
18-0.0232-0.0315-0.0423
19-0.0127-0.0321-0.0378
200.10870.10770.0420
21-0.0574-0.0676-0.0385
22-0.0415-0.0728-0.0431
23-0.1025-0.0895-0.0402
240.09360.11290.1256
250.00430.0222-0.0479
26-0.00560.0364-0.0341
270.07730.04010.0431
280.00690.02110.0578
29-0.0571-0.0481-0.0400
300.08240.04140.0415
310.09240.14670.0544
32-0.00360.0226-0.0325
330.08680.0663-0.0024
34-0.0149-0.0375-0.0429
35-0.0169-0.0367-0.0535
360.0164-0.01890.0084
370.02770.02910.0058
380.00280.07700.0162
39-0.0431-0.0604-0.0327
40-0.0599-0.0445-0.0481
410.08680.07650.0384
42-0.0087-0.0380-0.0413
430.08460.0666-0.0294
440.05970.06400.0058
450.00430.01050.0080
46-0.1003-0.0957-0.0610
470.05270.04890.0057
480.01650.0356-0.0443
49-0.0711-0.0505-0.0448
500.0004-0.0120-0.0512
510.06470.07650.0376
52-0.0234-0.0043-0.0396
530.01490.04270.0435
540.05690.04920.0012
550.06390.06560.1145
560.09520.10250.1649
57-0.0599-0.0484-0.0269
580.02250.0011-0.0372
590.00840.00320.0676
600.04170.04240.0841
61-0.0717-0.0395-0.0386
62-0.0373-0.0585-0.0365
630.05660.01380.0477
64-0.0084-0.0092-0.0370
650.10980.0176-0.0088
660.0102-0.0005-0.0274
670.08350.01370.0428
68-0.0361-0.0440-0.0447
690.04960.06560.0576
70-0.0846-0.0806-0.0602
710.01340.0022-0.0256
720.00260.01730.0200
730.09130.07540.0799
74-0.0156-0.0037-0.0448
750.07230.06710.0792
760.02350.01890.0228
77-0.0468-0.0744-0.0738
78-0.0644-0.0406-0.0605
790.01340.01500.0478
80-0.0908-0.0822-0.0441
81-0.0255-0.0484-0.0183
820.0076-0.0103-0.0399
83-0.0555-0.0476-0.0369
840.06330.00690.0621
850.08740.07130.1302
86-0.0367-0.0445-0.0396
87-0.00860.00170.0485
880.09640.09730.1271
890.02390.03460.0495
90-0.02900.02200.0205
91-0.0340-0.0445-0.0522
920.0161-0.00700.0184
93-0.01900.0135-0.0436
94-0.0807-0.0601-0.0679
95-0.0658-0.0403-0.0463
96-0.00770.0191-0.0078
97-0.0471-0.0414-0.0522
980.03470.07860.0324
990.0154-0.04680.0172
100-0.03780.0006-0.0145
1010.03470.0152-0.0226
1020.05570.07440.0330
103-0.0403-0.0138-0.0438
104-0.0476-0.0656-0.0578
105-0.0597-0.0604-0.0524
106-0.0667-0.0760-0.0342
107-0.0734-0.0942-0.0397
1080.19940.17590.2546
109-0.0315-0.0606-0.0426
110-0.0473-0.0637-0.0528
111-0.0510-0.0671-0.0366
112-0.0081-0.0088-0.0358
1130.09190.08640.0962
114-0.0346-0.0330-0.0407
1150.01620.0224-0.0095
116-0.0192-0.0042-0.0019
117-0.0041-0.00650.0582
118-0.0846-0.0676-0.0427
119-0.0119-0.0208-0.0456
120-0.0183-0.0492-0.0457
121-0.0018-0.0272-0.0035
122-0.0818-0.0848-0.0344
1230.02760.02500.0568
1240.13840.08840.0065
125-0.1288-0.1228-0.0431
1260.07620.05250.1178
127-0.0636-0.0848-0.0456
1280.03380.05780.0625
1290.09020.07600.0985
130-0.0695-0.0786-0.0446
1310.05970.0299-0.0271
1320.08070.03490.0282
1330.0159-0.0041-0.0019
1340.04990.02780.0925
135-0.0057-0.0313-0.0446
136-0.0153-0.00980.0226
1370.07840.07020.0679
1380.06780.0329-0.0244
1390.02310.07700.0275
140-0.0384-0.0315-0.0521
141-0.0153-0.0588-0.0165
142-0.0513-0.0570-0.0383
1430.02300.01230.0201
144-0.0992-0.1051-0.0462
1450.03040.00900.0026
1460.15240.10610.0856
1470.00590.0116-0.0159
1480.11600.17070.1765
1490.04960.11810.0805
1500.02900.06220.0940
1510.0339-0.03130.0772
152-0.0067-0.0276-0.0150
1530.00990.00800.0068
1540.07170.06300.0768
155-0.0038-0.02370.0103
1560.01670.0102-0.0170
157-0.0378-0.0565-0.0363
1580.02160.01190.0403
1590.0075-0.0269-0.0519
1600.08570.10560.1322
161-0.0591-0.0264-0.0762
162-0.0795-0.0900-0.0515
163-0.0885-0.0796-0.0494
164-0.0627-0.0276-0.0389
165-0.0294-0.0209-0.0060
166-0.0563-0.0242-0.0193
167-0.1008-0.1530-0.0562
1680.06530.06060.0634
1690.03000.0336-0.0260
170-0.0097-0.0042-0.0045
171-0.0403-0.0531-0.0390
1720.10640.07230.0920
1730.00330.00320.0660
174-0.0658-0.0575-0.0455
175-0.0457-0.0817-0.0735
1760.03750.04420.0706
177-0.0361-0.0385-0.0365
1780.0017-0.0173-0.0387
1790.08960.11450.1212
1800.24650.21440.1609
181-0.1053-0.0890-0.0253
182-0.0521-0.0432-0.0383
183-0.0543-0.0528-0.0556
184-0.0202-0.03930.0044
1850.04030.02250.0865
186-0.0790-0.0760-0.0535
1870.01790.01900.0738
1880.03920.00630.0777
1890.02020.03510.0621
1900.0127-0.01530.0522
1910.0225-0.00480.0777
1920.06020.08320.1218
1930.03350.09210.1112
1940.01190.0838-0.0298
195-0.0639-0.0630-0.0421
1960.08630.09570.1022
1970.06500.12330.0591
1980.01570.0101-0.0428
1990.05520.09730.0481
200-0.0647-0.0656-0.0513
2010.08180.01310.0435
202-0.0863-0.0601-0.0922
2030.11880.10200.0198
2040.11930.11810.1466
205-0.0162-0.0243-0.0064
2060.03310.0233-0.0101
2070.0155-0.02000.0105
208-0.0139-0.0098-0.0460
2090.0294-0.01590.0615
2100.16470.13420.2393
2110.0013-0.0325-0.0341
2120.0106-0.00600.0748
2130.0310-0.02190.0501
2140.23750.23310.2542
215-0.0448-0.0523-0.0373
216-0.06080.0142-0.0042
217-0.0902-0.0843-0.0610
218-0.0412-0.0177-0.0236
219-0.0639-0.0416-0.0577
220-0.0150-0.0708-0.0490
221-0.00970.0246-0.0291
222-0.0090-0.0325-0.0351
223-0.0395-0.0269-0.0356
224-0.0054-0.0226-0.0201
2250.08290.08840.1467
226-0.00080.0494-0.0105
2270.06890.05850.0245
228-0.01590.0265-0.0718
229-0.0028-0.02510.0085
230-0.02830.01460.0331
2310.04510.00630.0603
232-0.0317-0.03380.0059
233-0.0220-0.0105-0.0905
2340.20950.17070.1511
2350.11710.14150.1078
236-0.0109-0.0293-0.0381
237-0.0122-0.0267-0.0453
2380.03610.06870.0475
239-0.0262-0.0346-0.0458
2400.01900.03120.0457
2410.04540.01130.0542
242-0.0241-0.0238-0.0342
243-0.0767-0.1186-0.0570
244-0.0667-0.0682-0.0471
245-0.0619-0.0900-0.0421
246-0.0356-0.0874-0.0359
247-0.0300-0.0536-0.0373
248-0.0767-0.0640-0.0507
249-0.0127-0.0421-0.0073
Source: HUMANENESS_SMALL_VOICE_IDENTITY_R1_R2_2026-10-02_mediathek_0047.html · SHA-256 55f329d7c43b46e6…
g09EnglishAnime voice study

anime_000 · en · 6/50

Generated TTS prediction · Rank 1 qkvo · 5.4 s

It's just how things work, you know? (soft whistle) And honestly, I'm not sure if you can really control it.

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Doubt
z +2.64 / +3.43raw 1.24 / 1.61
Hope Enthusiasm Optimism
z +1.72 / +2.19raw 1.43 / 1.78
Contentment
z +0.80 / +1.38raw 0.41 / 0.71
Affection
z +0.73 / +1.31raw 0.42 / 0.75
Contemplation
z +0.75 / +1.07raw 0.53 / 0.72
Confusion
z +0.35 / +0.86raw 0.19 / 0.47
Teasing
z +0.80 / +0.34raw 0.44 / 0.18

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Vocal register (REGS)
z +2.27 / +2.33raw 4.08 / 4.15
Head resonance (R_HEAD)
z +2.23 / +2.28raw 3.98 / 4.02
Brightness (BRGT)
z +2.04 / +2.15raw 3.65 / 3.74
Storytelling style (S_STRY)
z +1.65 / +1.81raw 4.74 / 4.95
Cartoonish style (S_CART)
z +1.64 / +1.72raw 3.85 / 3.97
Conversational style (S_CONV)
z +1.54 / +1.35raw 4.62 / 4.37
Dramatic style (S_DRAM)
z +1.43 / +1.34raw 4.36 / 4.26

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z -0.61 / -0.70raw 1.34 / 1.19
Vocal-burst blend (0–10)z +0.30 / +0.43raw 7.51 / 7.86
Content enjoyment (CE)z +0.84 / +1.04raw 5.65 / 5.86
Production quality (PQ)z -0.01 / +0.24raw 7.04 / 7.26
Speech signal quality, calibratedz -0.34 / -0.17raw 3.32 / 3.40
Background quality, calibratedz +0.15 / +0.49raw 3.53 / 3.84
Overall quality, calibratedz -0.03 / +0.32raw 2.83 / 3.04
P.808 MOS estimatez -0.16 / +0.31raw 3.34 / 3.57
score content enjoymentz +0.72 / +0.77raw 4.73 / 4.74
score overall qualityz +0.01 / +0.15raw 2.65 / 2.73
Background Noisez -1.71 / -3.27raw 0.35 / -0.06
Tempo (TEMP)z +0.86 / +0.88raw 3.14 / 3.17
Arousal (AROU)z +0.11 / +0.19raw 2.49 / 2.58
Valence (VALN)z -0.12 / -0.19raw 2.19 / 2.11
Recording quality (RCQL)z +0.78 / +1.15raw 3.45 / 3.78
Background noise (BKGN)z +0.69 / +0.56raw 3.23 / 3.13
talking speedz +1.55 / +1.46raw 14.85 / 14.52
Characters per second (CPS): 20.00 raw · z +2.92

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8080.889
Identity · 250D0.6500.684

Vocal-burst regions

Tiny
Base
0 s5.4 s

Verified burst intervals are unavailable for this clip.

Tiny: 0 regions, count-head estimate 0.00; Base: 0 regions, count-head estimate 0.02.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.424+0.730.751+1.310.0070.570
1Amusementemo_Amusement0.247+0.280.110+0.13-0.0010.885
2Angeremo_Anger0.105+0.16-0.041-0.070.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.318+0.550.403+0.700.0050.572
4Aweemo_Awe-0.013-0.030.104+0.240.0000.427
5Bitternessemo_Bitterness0.002-0.030.034+0.050.0130.415
6Concentrationemo_Concentration0.368-0.360.237-0.540.6340.741
7Confusionemo_Confusion0.192+0.350.467+0.860.0010.540
8Contemplationemo_Contemplation0.531+0.750.722+1.070.0820.598
9Contemptemo_Contempt0.011+0.020.064+0.14-0.0000.464
10Contentmentemo_Contentment0.411+0.800.710+1.380.0000.514
11Disappointmentemo_Disappointment0.225+0.320.051+0.070.0000.700
12Disgustemo_Disgust-0.004-0.020.019+0.020.0070.489
13Distressemo_Distress0.058+0.07-0.116-0.140.0000.821
14Doubtemo_Doubt1.235+2.641.605+3.430.0060.466
15Elationemo_Elation0.461+0.620.486+0.650.0120.728
16Embarrassmentemo_Embarrassment0.068+0.06-0.019-0.140.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.079+0.080.004-0.000.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.045+0.030.006-0.040.0280.539
19Fearemo_Fear0.077+0.100.028+0.020.0150.610
20Helplessnessemo_Helplessness0.175+0.260.197+0.29-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism1.428+1.721.784+2.190.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.383+0.330.105-0.060.1490.705
23Infatuationemo_Infatuation-0.001+0.030.016+0.06-0.0150.533
24Interestemo_Interest1.938+0.392.024+0.481.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness0.014+0.070.010+0.06-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.042-0.040.040-0.040.0590.442
27Longingemo_Longing0.110+0.040.018-0.110.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.064+0.020.016-0.070.0550.534
29Painemo_Pain-0.003-0.00-0.037-0.06-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.399+0.620.507+0.800.0170.614
31Prideemo_Pride-0.009-0.020.023+0.040.0020.509
32Reliefemo_Relief0.227+0.280.423+0.570.0330.681
33Sadnessemo_Sadness0.056+0.070.077+0.100.0000.759
34Sexual Lustemo_Sexual_Lust0.078+0.10-0.008-0.080.0300.494
35Shameemo_Shame-0.010-0.040.006-0.010.0130.510
36Sournessemo_Sourness0.148+0.350.001+0.02-0.0050.439
37Teasingemo_Teasing0.438+0.800.183+0.340.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.072+0.040.203+0.230.0410.697
39Triumphemo_Triumph-0.018-0.020.017+0.05-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg2.002-0.792.216-0.622.9841.238
41Arousal (AROU)vn_AROU_reg2.486+0.112.579+0.192.3591.147
42Arousal shift (ARSH)vn_ARSH_reg2.697+0.142.686+0.122.5800.867
43Vocal attack (ATCK)vn_ATCK_reg1.432-0.561.360-0.622.1621.295
44Background noise (BKGN)vn_BKGN_reg3.228+0.693.131+0.562.7030.761
45Brightness (BRGT)vn_BRGT_reg3.648+2.043.737+2.151.9910.813
46Speech chunking (CHNK)vn_CHNK_reg2.612+0.622.617+0.621.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg2.324+0.322.532+0.542.0180.945
48Cognitive load (COGL)vn_COGL_reg1.845-0.231.583-0.482.0941.061
49Dynamic arc (DARC)vn_DARC_reg2.538+0.672.538+0.671.6151.380
50Disfluency (DFLU)vn_DFLU_reg2.216-0.232.118-0.312.4881.184
51Emphasis (EMPH)vn_EMPH_reg3.068+0.303.289+0.452.6071.524
52Esthetics (ESTH)vn_ESTH_reg3.444+1.333.377+1.252.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg-0.151-0.74-0.169-0.790.0860.322
54Focus (FOCS)vn_FOCS_reg3.386+0.953.386+0.952.3441.102
55Fullness (FULL)vn_FULL_reg1.399-0.431.627-0.171.7780.889
56Perceived gender (GEND)vn_GEND_reg0.435-2.280.535-2.224.0551.588
57Harmonicity (HARM)vn_HARM_reg1.404-1.291.353-1.352.4800.836
58Metallic character (METL)vn_METL_reg0.966-0.680.902-0.751.6210.958
59Pitch range (RANG)vn_RANG_reg4.062+1.284.055+1.272.9140.897
60Recording quality (RCQL)vn_RCQL_reg3.448+0.783.778+1.152.7540.888
61Vocal register (REGS)vn_REGS_reg4.082+2.274.151+2.331.1981.268
62Respiration (RESP)vn_RESP_reg3.578+0.853.402+0.682.6801.056
63Roughness (ROUG)vn_ROUG_reg1.365-0.751.423-0.692.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg1.060-1.271.013-1.332.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg3.984+2.234.022+2.282.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg2.914+1.003.042+1.131.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg1.452+0.021.358-0.091.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg3.601+0.773.291+0.482.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg2.705+0.702.504+0.432.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg1.878-0.241.834-0.292.0700.803
71Smoothness (SMTH)vn_SMTH_reg2.291+0.142.315+0.162.1431.093
72Stance (STNC)vn_STNC_reg1.595-0.911.604-0.902.6331.137
73Speech structure (STRU)vn_STRU_reg2.961+0.862.938+0.842.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg1.452+0.061.612+0.171.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg0.177-0.870.323-0.731.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg3.851+1.643.969+1.721.5601.401
77Casual style (S_CASU)vn_S_CASU_reg3.819+0.023.947+0.123.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg4.624+1.544.369+1.352.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg4.364+1.434.257+1.342.7211.148
80Formal style (S_FORM)vn_S_FORM_reg0.938-0.250.948-0.241.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg2.576-0.192.769-0.042.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg3.011+1.183.090+1.231.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.221-0.330.319-0.160.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg3.914+1.193.965+1.232.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg1.730+0.401.763+0.431.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg4.737+1.654.955+1.812.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg1.123+0.391.304+0.540.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg3.703+1.303.391+1.111.5971.618
89Tempo (TEMP)vn_TEMP_reg3.137+0.863.170+0.882.1171.190
90Tension (TENS)vn_TENS_reg1.690+0.151.897+0.401.5590.852
91Valence (VALN)vn_VALN_reg2.193-0.122.111-0.192.3281.118
92Valence shift (VALS)vn_VALS_reg2.884+0.342.674+0.142.5181.079
93Velocity flux (VFLX)vn_VFLX_reg3.041+0.652.961+0.582.2561.208
94Volatility (VOLT)vn_VOLT_reg3.424+0.953.397+0.922.5860.886
95Vulnerability (VULN)vn_VULN_reg3.533+0.903.622+0.992.6480.984
96Warmth (WARM)vn_WARM_reg2.072-0.562.216-0.402.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_61.338-0.611.194-0.702.3321.621
98Vocal-burst blend (0–10)blend_0_107.509+0.307.865+0.436.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.976-0.350.987-0.321.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.044-0.703.109-0.483.2500.293
101Arousaleiv_extra:Arousal1.602+0.721.583+0.681.2890.435
102Authenticityeiv_extra:Authenticity3.008+2.233.141+3.102.6670.153
103Background Noiseeiv_extra:Background_Noise0.354-1.71-0.056-3.270.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.456+1.381.337+0.981.0400.303
105Gendereiv_extra:Gender1.185+1.811.325+1.95-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.378-1.801.210-2.471.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive2.188+0.702.214+0.761.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.544+1.202.731+1.772.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous1.142+0.621.248+0.940.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.662+1.190.902+1.940.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.390+0.040.414+0.120.3760.324
112Valenceeiv_extra:Valence0.260+0.670.438+1.140.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.235-1.100.928-1.801.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold0.491+1.470.965+3.040.0510.300
115score background qualityeiv_extra:score_background_quality3.379+0.113.456+0.223.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.726+0.724.742+0.774.5070.303
117score overall qualityeiv_extra:score_overall_quality2.655+0.012.727+0.152.6520.507
118score speech qualityeiv_extra:score_speech_quality1.893+0.581.901+0.641.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.649+0.845.856+1.044.7991.018
120Content usefulness (CU)audiobox:CU6.401+0.756.667+1.025.6491.001
121Production complexity (PC)audiobox:PC1.820-0.321.587-0.462.3311.617
122Production quality (PQ)audiobox:PQ7.039-0.017.256+0.247.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw3.614-0.433.766-0.203.9000.669
124Background quality, rawdnsmos:BAK_raw3.601+0.353.920+0.633.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.111+0.033.395+0.383.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.320-0.343.403-0.173.4880.501
127Background quality, calibrateddnsmos:BAK3.531+0.153.840+0.493.3890.928
128Overall quality, calibrateddnsmos:OVRL2.825-0.033.037+0.322.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.339-0.163.575+0.313.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p-0.005-0.010.015+0.020.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection1.544+1.081.212+0.480.9470.554
132Agevoiceclap_attribute:Age2.436-1.452.724-0.923.2290.547
133Amusementvoiceclap_attribute:Amusement0.641+0.430.685+0.510.3920.572
134Angervoiceclap_attribute:Anger0.708-0.560.514-0.950.9920.506
135Arousalvoiceclap_attribute:Arousal1.917+0.801.595+0.281.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise1.482+2.031.654+2.390.5090.479
137Authenticityvoiceclap_attribute:Authenticity3.315+3.263.347+3.452.7700.167
138Awevoiceclap_attribute:Awe0.714+0.680.999+1.770.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.228-1.290.156-1.560.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.243-1.480.204-1.650.5920.236
141Concentrationvoiceclap_attribute:Concentration0.547-1.260.646-1.071.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.370+0.641.500+0.921.0690.468
143Confusionvoiceclap_attribute:Confusion1.064+1.260.850+0.620.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.469-0.770.512-0.660.7800.406
145Contemptvoiceclap_attribute:Contempt0.399-0.920.247-1.360.7090.339
146Contentmentvoiceclap_attribute:Contentment0.646+0.190.667+0.260.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.914+0.530.800+0.140.7600.291
148Disgustvoiceclap_attribute:Disgust0.638+0.060.589-0.170.6250.207
149Distressvoiceclap_attribute:Distress1.130+0.581.148+0.630.9270.350
150Doubtvoiceclap_attribute:Doubt0.791+1.360.853+1.690.5350.188
151Elationvoiceclap_attribute:Elation0.519-0.050.686+0.370.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.454+0.470.543+0.750.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.488-0.660.414-0.940.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.233-0.880.223-0.910.4620.261
155Fearvoiceclap_attribute:Fear0.925+1.450.925+1.450.6180.212
156Gendervoiceclap_attribute:Gender1.380+1.521.356+1.50-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness1.032+1.390.872+0.800.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.122-1.921.175-1.781.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism1.084+0.961.370+1.680.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability0.994-0.110.889-0.301.0520.541
161Infatuationvoiceclap_attribute:Infatuation1.015+1.220.963+1.070.6080.332
162Interestvoiceclap_attribute:Interest1.698+0.231.827+0.521.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.200-0.650.111-0.840.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy1.029+2.071.066+2.220.5230.245
165Longingvoiceclap_attribute:Longing0.873+1.520.847+1.410.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.500-0.960.362-1.360.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive2.031+0.432.357+0.981.7750.594
168Painvoiceclap_attribute:Pain0.571+0.870.411+0.140.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.729+0.750.604+0.340.5030.301
170Pridevoiceclap_attribute:Pride0.329-1.290.318-1.320.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.725+1.262.838+1.602.3130.328
172Reliefvoiceclap_attribute:Relief1.195+0.411.213+0.461.0250.415
173Sadnessvoiceclap_attribute:Sadness0.992+0.771.002+0.800.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.758-0.080.948+0.260.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.361-0.650.441-0.460.6350.419
176Shamevoiceclap_attribute:Shame0.794+0.940.901+1.320.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.757+1.660.579+1.170.1600.360
178Sournessvoiceclap_attribute:Sourness0.794+0.300.535-0.590.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.152-1.310.166-1.280.6230.359
180Teasingvoiceclap_attribute:Teasing0.832+0.770.751+0.560.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.586+0.381.298-0.161.3840.529
182Triumphvoiceclap_attribute:Triumph0.333-1.070.513-0.570.7160.359
183Valencevoiceclap_attribute:Valence-0.105-0.150.392+0.63-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached0.532-1.990.454-2.131.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold0.236+0.760.461+1.32-0.0660.400
186durationvoiceclap_attribute:duration7.479-0.546.984-0.6210.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.511+0.533.462+0.443.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.792+0.934.751+0.654.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.655+0.052.704+0.172.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.964+1.051.961+1.021.8910.069
191talking speedvoiceclap_attribute:talking_speed14.846+1.5514.524+1.469.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.0064-0.0146-0.0671
10.16840.12170.0802
20.10040.10790.1346
3-0.0948-0.1142-0.0776
4-0.0427-0.0479-0.0720
5-0.1316-0.1119-0.0969
60.09790.05420.0241
70.17590.17070.2299
8-0.01610.0231-0.0067
9-0.0335-0.07500.0759
10-0.0437-0.04670.0359
11-0.0440-0.0761-0.0895
12-0.0885-0.0859-0.0770
13-0.0717-0.0314-0.0451
140.15340.15920.1450
15-0.01810.04330.0380
16-0.1066-0.1021-0.0948
17-0.0655-0.0643-0.0796
18-0.0811-0.0666-0.0087
19-0.0577-0.0859-0.0724
20-0.0229-0.0606-0.0767
21-0.01030.0012-0.0035
22-0.0948-0.0482-0.0615
23-0.1659-0.0992-0.0906
24-0.0189-0.00150.0456
25-0.0811-0.0744-0.0673
26-0.0611-0.0649-0.0718
27-0.0345-0.0389-0.0714
28-0.0748-0.0854-0.0844
290.10540.04270.0769
300.07610.0395-0.0533
31-0.0549-0.0600-0.0308
320.22450.26760.1733
330.0129-0.0267-0.0347
34-0.0689-0.0756-0.0687
350.08540.09810.0962
36-0.0948-0.0600-0.0823
37-0.0505-0.0415-0.0906
380.08850.15340.1613
39-0.1048-0.0470-0.0772
400.06270.08020.0440
41-0.0004-0.0464-0.0521
42-0.0733-0.0738-0.0772
43-0.0288-0.0120-0.0149
440.0293-0.00610.0014
45-0.0265-0.0796-0.0461
460.19950.13090.0773
47-0.1191-0.1246-0.0774
48-0.0892-0.0906-0.0748
49-0.0564-0.0317-0.0551
500.05580.0093-0.0037
510.01380.09000.0914
52-0.0552-0.0268-0.0421
53-0.0543-0.0718-0.0781
54-0.0904-0.1021-0.0857
55-0.0357-0.0735-0.0103
56-0.0733-0.0934-0.0726
570.0030-0.0092-0.0427
58-0.04210.01020.0598
590.07110.05130.0828
600.05210.07090.0582
610.17340.16380.1572
62-0.03090.08710.0773
630.0120-0.0232-0.0457
64-0.0081-0.0018-0.0233
65-0.0608-0.0790-0.0624
660.12850.10320.1564
67-0.0486-0.0678-0.0739
680.07200.02810.0307
690.03150.02390.0657
700.05110.07610.0361
710.29680.23420.1873
72-0.03020.03240.0271
730.0231-0.0039-0.0165
74-0.03650.02350.1207
750.24940.22960.1770
760.12780.12570.1493
77-0.0555-0.0842-0.0699
78-0.0720-0.0718-0.0681
79-0.01100.0510-0.0664
80-0.0455-0.0424-0.0389
810.12350.18340.2500
82-0.1029-0.0666-0.0732
830.17460.17420.2549
84-0.1035-0.1096-0.0949
850.04930.08820.0621
86-0.0148-0.02420.0650
87-0.0539-0.0738-0.0884
88-0.0168-0.00440.0422
89-0.0465-0.0767-0.0703
900.04240.10610.0088
910.0144-0.00910.0128
92-0.0214-0.0212-0.0074
93-0.0730-0.0975-0.0596
94-0.0131-0.0257-0.0558
95-0.0733-0.0378-0.0822
96-0.1147-0.1523-0.0806
970.12530.14770.1438
98-0.0490-0.0643-0.0432
990.11660.14070.1155
1000.14650.16150.1823
1010.02400.0040-0.0740
102-0.01570.01510.0938
103-0.0773-0.0796-0.0808
1040.09670.09000.0065
105-0.0670-0.0750-0.0739
1060.04270.01060.0508
107-0.0546-0.0669-0.0476
108-0.0829-0.0888-0.0831
109-0.0848-0.0534-0.0481
1100.04430.0320-0.0097
1110.03240.04820.1080
112-0.0885-0.0767-0.0754
1130.05110.0337-0.0024
1140.07480.08310.0854
1150.13100.17070.1731
1160.10980.13610.0777
117-0.01620.00340.0281
118-0.0773-0.0854-0.0745
1190.08230.0021-0.0740
1200.05240.10670.1201
121-0.05330.07840.0955
122-0.0683-0.0836-0.0617
1230.01720.0130-0.0052
1240.05710.01720.1062
125-0.0465-0.0663-0.0763
126-0.1172-0.0675-0.0660
1270.14030.05220.0012
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.00750.04840.0126
1-0.1033-0.0524-0.0381
2-0.04440.0105-0.0485
30.0152-0.0526-0.0648
40.11230.07580.1460
5-0.0341-0.0361-0.0412
60.01220.06290.0048
70.06930.10220.2075
80.04800.03880.0274
90.03630.05240.0920
10-0.1084-0.0332-0.0414
11-0.0645-0.0148-0.0277
12-0.0436-0.03060.0285
130.05250.13920.1567
14-0.0441-0.0172-0.0293
150.12620.11870.1577
16-0.00270.0161-0.0637
170.0425-0.0271-0.0473
180.14860.10710.1893
19-0.0933-0.0830-0.0625
200.02460.06740.0450
21-0.0165-0.0180-0.0366
22-0.01310.02260.0367
23-0.0771-0.0839-0.0381
240.0293-0.0091-0.0337
250.03380.0515-0.0044
26-0.00120.0294-0.0435
27-0.0112-0.0549-0.0320
28-0.0032-0.0034-0.0427
29-0.0036-0.01530.0829
300.08830.01800.0858
310.15860.19630.2118
320.16420.17490.0387
330.05500.0569-0.0106
340.06090.15350.0357
350.05700.02590.0014
360.06030.06780.0999
37-0.0871-0.0439-0.0272
38-0.0860-0.0897-0.0547
39-0.0413-0.1098-0.0587
400.0316-0.00340.0023
410.01580.00580.0216
420.01080.0317-0.0351
430.06280.0598-0.0305
440.07150.05710.0096
45-0.06810.0299-0.0047
46-0.1139-0.1410-0.0579
470.0288-0.0166-0.0516
480.0053-0.03190.0270
49-0.00030.01140.0816
500.0152-0.0399-0.0055
51-0.01490.0448-0.0082
520.03560.0224-0.0383
53-0.0229-0.0727-0.0381
540.0776-0.0059-0.0033
550.08880.0253-0.0503
56-0.0334-0.0377-0.0140
570.05030.02400.0641
58-0.06540.0048-0.0378
590.02090.11780.0303
600.0037-0.0128-0.0259
61-0.0436-0.0257-0.0366
620.0175-0.0758-0.0346
630.06400.06740.1023
64-0.0514-0.0718-0.0439
650.0256-0.00700.0744
660.09160.12850.2050
67-0.0888-0.1579-0.0414
680.01230.0406-0.0424
690.07210.09010.0557
700.04830.01290.0704
71-0.0125-0.0381-0.0413
720.00490.06420.0294
730.06870.05290.0214
74-0.0910-0.0883-0.0425
750.13680.12760.1153
760.09720.11420.0947
77-0.0575-0.0110-0.0125
780.04410.0366-0.0013
79-0.0363-0.0736-0.0422
80-0.00040.00390.0397
81-0.00050.00530.0032
82-0.0270-0.0651-0.0378
83-0.0216-0.0021-0.0350
840.02610.07090.0538
850.05950.0066-0.0117
860.04380.0282-0.0376
87-0.0793-0.0257-0.0336
880.08320.05930.1775
890.05640.01360.1090
90-0.04130.0011-0.0417
910.06540.05310.1204
920.02630.02110.0379
930.04440.0560-0.0042
94-0.0026-0.0446-0.0681
950.18990.16510.1515
960.03480.09460.1537
970.01500.06290.1204
98-0.0219-0.02750.0429
99-0.06670.02030.0056
1000.10110.05690.0975
1010.10500.0576-0.0398
102-0.0452-0.1196-0.0470
1030.02070.04100.0071
104-0.0338-0.0781-0.0548
105-0.0178-0.0321-0.0046
106-0.0511-0.0404-0.0324
107-0.00440.0279-0.0326
1080.02080.0020-0.0313
109-0.1000-0.0529-0.0405
110-0.0586-0.0589-0.0501
1110.22120.11870.1652
112-0.03180.03970.0487
113-0.0667-0.0892-0.0717
114-0.0832-0.0732-0.0587
115-0.0169-0.0393-0.0298
1160.10670.11870.0602
1170.0320-0.05440.0044
118-0.04890.04970.0069
1190.03550.05690.0266
120-0.0143-0.0074-0.0536
1210.09890.07630.0442
122-0.0550-0.0435-0.0397
1230.01290.04770.0324
1240.00060.06070.0019
1250.12790.08340.0461
1260.04110.08920.0640
127-0.0304-0.03320.0164
128-0.0514-0.0709-0.0524
129-0.0228-0.0482-0.0446
130-0.02860.0161-0.0423
131-0.0251-0.0377-0.0368
1320.04020.08740.0973
133-0.01350.0156-0.0416
1340.10330.0772-0.0750
135-0.0592-0.0611-0.0423
136-0.04720.0168-0.0451
137-0.0187-0.01760.0033
138-0.0232-0.0359-0.0371
1390.0148-0.0417-0.0430
140-0.0214-0.0598-0.0325
1410.06230.03930.0246
142-0.0799-0.0102-0.0363
1430.03970.03520.0399
144-0.0352-0.0213-0.0145
145-0.0581-0.0388-0.0535
1460.12960.08570.0362
147-0.0469-0.0660-0.0374
148-0.0073-0.0076-0.0464
1490.07370.12400.0504
150-0.01080.05110.0044
1510.0550-0.00780.0545
152-0.0163-0.0504-0.0467
153-0.01290.0292-0.0663
1540.0592-0.0114-0.0193
1550.02640.06380.0345
156-0.0427-0.0252-0.0421
1570.12230.08880.0271
1580.08040.03210.0484
1590.13570.14810.1832
160-0.10450.0359-0.0399
161-0.01280.0339-0.0253
162-0.1430-0.0799-0.0488
1630.0276-0.0012-0.0468
1640.0411-0.0139-0.0370
1650.04750.10260.1180
1660.0205-0.0491-0.0490
1670.01120.0143-0.0505
168-0.0279-0.0249-0.0331
1690.0179-0.0647-0.0480
170-0.0093-0.0897-0.0106
1710.07260.0959-0.0146
1720.01950.0448-0.0217
1730.11620.06600.0756
174-0.02950.02480.0421
175-0.0290-0.0433-0.0697
176-0.0441-0.0763-0.0358
1770.0193-0.0243-0.0347
1780.04860.05580.0030
1790.0276-0.0005-0.0518
1800.06620.10480.0522
181-0.0480-0.1285-0.0600
182-0.0656-0.0301-0.0315
183-0.0614-0.0558-0.0848
1840.03230.0153-0.0427
185-0.00720.00530.0440
186-0.0408-0.0462-0.0507
1870.0296-0.00110.0907
188-0.02700.02920.0316
1890.03270.01410.0021
1900.10780.04480.0012
191-0.00970.0088-0.0477
1920.02900.0602-0.0448
193-0.0419-0.0292-0.0382
194-0.0427-0.01090.0006
195-0.0827-0.0326-0.0399
196-0.0107-0.00470.0179
1970.00180.02940.0336
198-0.0707-0.0576-0.0406
199-0.0274-0.0480-0.0462
200-0.0422-0.1004-0.0486
2010.02640.03880.0900
2020.00890.07410.0306
203-0.0229-0.0785-0.0525
2040.02930.01380.0815
2050.07650.03610.0627
2060.13570.13300.0848
207-0.00650.00200.0960
208-0.00570.0262-0.0436
209-0.0303-0.0424-0.0297
210-0.0138-0.0377-0.0547
211-0.05390.0014-0.0050
212-0.01370.03260.0995
2130.16090.15170.0740
2140.03990.02160.0724
2150.07090.07230.0922
2160.20000.17760.1321
2170.03630.01770.0741
218-0.1257-0.1374-0.0488
219-0.02890.01580.0088
220-0.0910-0.0941-0.0464
221-0.0021-0.0359-0.0365
222-0.0209-0.0328-0.0425
223-0.02670.03280.0563
2240.01380.02580.1022
225-0.0228-0.0115-0.0350
2260.10610.0232-0.0044
2270.0447-0.02920.1427
2280.00320.0262-0.0624
2290.0514-0.01580.0108
230-0.0411-0.0180-0.0428
2310.02560.02400.0126
2320.00920.05260.0287
2330.05890.05040.0609
2340.03660.0262-0.0326
2350.09830.10980.0297
236-0.0531-0.0448-0.0361
237-0.0230-0.0651-0.0429
238-0.01360.00210.0286
2390.02670.0417-0.0435
240-0.0698-0.0446-0.0429
2410.0760-0.01620.0862
242-0.0589-0.0674-0.0619
243-0.0748-0.0200-0.0540
2440.04920.04130.0299
245-0.00070.03080.0620
246-0.03990.0002-0.0341
247-0.00410.0117-0.0354
248-0.0978-0.0616-0.0481
2490.03940.0625-0.0203
Source: HUMANENESS_SMALL_VOICE_IDENTITY_R1_R2_2026-10-02_anime_000.html · SHA-256 f301c97c0aa149c3…
g10EnglishAnime voice study

anime_000 · en · 25/50

Generated TTS prediction · Rank 2 qv · 9.6 s

This long development points to serious damage in the reproductive system or maybe some infection setting in. (low mumble) We need to figure out what is causing this difficult situation right away.

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Concentration
z +2.13 / +2.48raw 2.21 / 2.47
Fear
z +2.17 / +2.40raw 1.34 / 1.48
Doubt
z +2.10 / +1.60raw 0.99 / 0.75
Contemplation
z +1.75 / +1.37raw 1.13 / 0.90
Distress
z +1.17 / +1.11raw 0.96 / 0.91
Impatience and Irritability
z +1.05 / +1.14raw 0.89 / 0.95
Confusion
z +1.07 / +1.09raw 0.58 / 0.59

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Vocal register (REGS)
z +2.25 / +2.20raw 4.05 / 3.98
Narrator style (S_NARR)
z +1.83 / +2.01raw 3.95 / 4.21
Head resonance (R_HEAD)
z +1.74 / +1.75raw 3.58 / 3.59
Whispered talking style (S_WHIS)
z +1.70 / +1.72raw 4.35 / 4.38
Tempo (TEMP)
z +1.36 / +1.46raw 3.74 / 3.86
Esthetics (ESTH)
z +1.30 / +1.34raw 3.42 / 3.45
Storytelling style (S_STRY)
z +1.15 / +1.25raw 4.05 / 4.19

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z -0.71 / -0.76raw 1.18 / 1.09
Vocal-burst blend (0–10)z -0.78 / -0.74raw 4.54 / 4.65
Content enjoyment (CE)z +0.78 / +1.15raw 5.59 / 5.97
Production quality (PQ)z +0.18 / -0.00raw 7.20 / 7.04
Speech signal quality, calibratedz +0.16 / +0.25raw 3.57 / 3.61
Background quality, calibratedz +0.56 / +0.59raw 3.90 / 3.93
Overall quality, calibratedz +0.52 / +0.68raw 3.16 / 3.25
P.808 MOS estimatez +0.47 / +0.45raw 3.66 / 3.65
score content enjoymentz +0.58 / +0.30raw 4.68 / 4.60
score overall qualityz +0.39 / +0.36raw 2.85 / 2.83
Background Noisez -1.26 / -1.11raw 0.47 / 0.52
Tempo (TEMP)z +1.36 / +1.46raw 3.74 / 3.86
Arousal (AROU)z +0.23 / +0.06raw 2.62 / 2.43
Valence (VALN)z -0.22 / +0.04raw 2.08 / 2.37
Recording quality (RCQL)z +0.75 / +0.54raw 3.42 / 3.23
Background noise (BKGN)z +0.95 / +0.83raw 3.43 / 3.34
talking speedz +1.94 / +1.99raw 16.31 / 16.49
Characters per second (CPS): 20.62 raw · z +3.09

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.9050.914
Identity · 250D0.6730.712

Vocal-burst regions

Tiny
Base
0 s9.6 s

Verified burst intervals are unavailable for this clip.

Tiny: 1 regions, count-head estimate 1.19; Base: 1 regions, count-head estimate 1.08.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
Tiny10.020.220.20
Base10.020.220.20
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.049+0.070.044+0.060.0070.570
1Amusementemo_Amusement-0.031-0.03-0.085-0.10-0.0010.885
2Angeremo_Anger0.490+0.790.438+0.710.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.061+0.100.045+0.070.0050.572
4Aweemo_Awe0.014+0.030.012+0.030.0000.427
5Bitternessemo_Bitterness-0.016-0.07-0.005-0.040.0130.415
6Concentrationemo_Concentration2.212+2.132.474+2.480.6340.741
7Confusionemo_Confusion0.579+1.070.589+1.090.0010.540
8Contemplationemo_Contemplation1.131+1.750.904+1.370.0820.598
9Contemptemo_Contempt0.010+0.020.001+0.00-0.0000.464
10Contentmentemo_Contentment-0.043-0.08-0.008-0.020.0000.514
11Disappointmentemo_Disappointment0.455+0.650.445+0.640.0000.700
12Disgustemo_Disgust-0.020-0.060.012+0.010.0070.489
13Distressemo_Distress0.963+1.170.912+1.110.0000.821
14Doubtemo_Doubt0.986+2.100.752+1.600.0060.466
15Elationemo_Elation-0.042-0.07-0.014-0.040.0120.728
16Embarrassmentemo_Embarrassment0.021-0.05-0.008-0.120.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.386+0.420.118+0.120.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.068+0.070.022-0.010.0280.539
19Fearemo_Fear1.340+2.171.478+2.400.0150.610
20Helplessnessemo_Helplessness0.676+0.980.593+0.86-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.326+0.270.213+0.130.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.888+1.050.950+1.140.1490.705
23Infatuationemo_Infatuation0.021+0.07-0.018-0.01-0.0150.533
24Interestemo_Interest1.992+0.442.199+0.671.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness-0.043-0.03-0.039-0.03-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.027-0.070.035-0.050.0590.442
27Longingemo_Longing0.005-0.13-0.017-0.160.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.062+0.010.052-0.010.0550.534
29Painemo_Pain-0.001+0.000.014+0.03-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.008-0.010.001-0.030.0170.614
31Prideemo_Pride-0.025-0.050.032+0.060.0020.509
32Reliefemo_Relief0.070+0.050.156+0.180.0330.681
33Sadnessemo_Sadness0.398+0.520.441+0.580.0000.759
34Sexual Lustemo_Sexual_Lust-0.068-0.20-0.041-0.140.0300.494
35Shameemo_Shame0.022+0.02-0.015-0.050.0130.510
36Sournessemo_Sourness0.045+0.12-0.048-0.10-0.0050.439
37Teasingemo_Teasing-0.028-0.05-0.018-0.030.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude-0.005-0.070.039-0.000.0410.697
39Triumphemo_Triumph-0.019-0.02-0.009-0.00-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg2.873-0.092.667-0.262.9841.238
41Arousal (AROU)vn_AROU_reg2.625+0.232.434+0.062.3591.147
42Arousal shift (ARSH)vn_ARSH_reg2.403-0.202.415-0.192.5800.867
43Vocal attack (ATCK)vn_ATCK_reg1.568-0.461.630-0.412.1621.295
44Background noise (BKGN)vn_BKGN_reg3.429+0.953.338+0.832.7030.761
45Brightness (BRGT)vn_BRGT_reg2.622+0.782.429+0.541.9910.813
46Speech chunking (CHNK)vn_CHNK_reg3.075+0.992.953+0.891.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg3.148+1.203.035+1.082.0180.945
48Cognitive load (COGL)vn_COGL_reg1.733-0.341.482-0.582.0941.061
49Dynamic arc (DARC)vn_DARC_reg1.632+0.011.563-0.041.6151.380
50Disfluency (DFLU)vn_DFLU_reg1.629-0.731.343-0.972.4881.184
51Emphasis (EMPH)vn_EMPH_reg2.721+0.072.779+0.112.6071.524
52Esthetics (ESTH)vn_ESTH_reg3.420+1.303.450+1.342.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.141+0.170.183+0.300.0860.322
54Focus (FOCS)vn_FOCS_reg3.655+1.193.620+1.162.3441.102
55Fullness (FULL)vn_FULL_reg2.000+0.251.744-0.041.7780.889
56Perceived gender (GEND)vn_GEND_reg0.684-2.120.373-2.324.0551.588
57Harmonicity (HARM)vn_HARM_reg1.637-1.011.679-0.962.4800.836
58Metallic character (METL)vn_METL_reg1.463-0.161.372-0.261.6210.958
59Pitch range (RANG)vn_RANG_reg3.522+0.683.755+0.942.9140.897
60Recording quality (RCQL)vn_RCQL_reg3.423+0.753.230+0.542.7540.888
61Vocal register (REGS)vn_REGS_reg4.053+2.253.984+2.201.1981.268
62Respiration (RESP)vn_RESP_reg3.411+0.693.145+0.442.6801.056
63Roughness (ROUG)vn_ROUG_reg1.210-0.931.241-0.892.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg1.338-0.941.229-1.072.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg3.578+1.743.591+1.752.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg2.786+0.872.590+0.681.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg1.808+0.441.525+0.111.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg3.980+1.143.757+0.922.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg2.321+0.192.117-0.092.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg2.232+0.202.154+0.102.0700.803
71Smoothness (SMTH)vn_SMTH_reg2.966+0.752.862+0.662.1431.093
72Stance (STNC)vn_STNC_reg2.278-0.312.171-0.412.6331.137
73Speech structure (STRU)vn_STRU_reg3.108+1.013.146+1.052.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg2.621+0.892.734+0.971.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg1.318+0.221.227+0.131.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg2.867+0.932.776+0.871.5601.401
77Casual style (S_CASU)vn_S_CASU_reg3.496-0.233.322-0.373.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg3.492+0.683.365+0.582.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg3.518+0.693.638+0.802.7211.148
80Formal style (S_FORM)vn_S_FORM_reg2.203+0.992.179+0.961.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg3.079+0.203.046+0.172.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg3.953+1.834.211+2.011.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.343-0.110.360-0.080.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg3.010+0.503.299+0.722.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg1.385+0.131.013-0.161.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg4.051+1.154.193+1.252.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg1.180+0.441.171+0.430.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg4.352+1.704.377+1.721.5971.618
89Tempo (TEMP)vn_TEMP_reg3.738+1.363.860+1.462.1171.190
90Tension (TENS)vn_TENS_reg2.141+0.681.921+0.431.5590.852
91Valence (VALN)vn_VALN_reg2.077-0.222.369+0.042.3281.118
92Valence shift (VALS)vn_VALS_reg2.337-0.172.321-0.182.5181.079
93Velocity flux (VFLX)vn_VFLX_reg2.726+0.392.961+0.582.2561.208
94Volatility (VOLT)vn_VOLT_reg2.462-0.142.602+0.022.5860.886
95Vulnerability (VULN)vn_VULN_reg2.921+0.282.861+0.222.6480.984
96Warmth (WARM)vn_WARM_reg2.239-0.372.271-0.342.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_61.181-0.711.093-0.762.3321.621
98Vocal-burst blend (0–10)blend_0_104.542-0.784.648-0.746.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.671-1.260.661-1.291.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.238-0.043.294+0.153.2500.293
101Arousaleiv_extra:Arousal1.490+0.461.562+0.631.2890.435
102Authenticityeiv_extra:Authenticity2.946+1.822.878+1.382.6670.153
103Background Noiseeiv_extra:Background_Noise0.474-1.260.515-1.110.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.446+1.341.367+1.081.0400.303
105Gendereiv_extra:Gender0.795+1.420.666+1.29-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.747-0.321.680-0.591.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive1.886+0.061.922+0.131.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.588+1.342.499+1.072.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous0.449-1.410.609-0.940.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.456+0.550.469+0.590.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.344-0.100.395+0.060.3760.324
112Valenceeiv_extra:Valence-0.185-0.50-0.163-0.440.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.844+0.301.834+0.271.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold-0.141-0.64-0.017-0.230.0510.300
115score background qualityeiv_extra:score_background_quality3.502+0.293.418+0.173.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.684+0.584.599+0.304.5070.303
117score overall qualityeiv_extra:score_overall_quality2.850+0.392.832+0.362.6520.507
118score speech qualityeiv_extra:score_speech_quality1.877+0.471.833+0.181.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.594+0.785.967+1.154.7991.018
120Content usefulness (CU)audiobox:CU6.573+0.926.721+1.075.6491.001
121Production complexity (PC)audiobox:PC1.454-0.541.404-0.572.3311.617
122Production quality (PQ)audiobox:PQ7.201+0.187.043-0.007.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw4.025+0.194.118+0.333.9000.669
124Background quality, rawdnsmos:BAK_raw4.030+0.734.065+0.763.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.564+0.593.726+0.793.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.569+0.163.614+0.253.4880.501
127Background quality, calibrateddnsmos:BAK3.905+0.563.934+0.593.3890.928
128Overall quality, calibrateddnsmos:OVRL3.157+0.523.252+0.682.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.655+0.473.645+0.453.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.784+1.210.732+1.130.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.956+0.021.040+0.170.9470.554
132Agevoiceclap_attribute:Age2.684-1.002.800-0.793.2290.547
133Amusementvoiceclap_attribute:Amusement0.274-0.210.557+0.290.3920.572
134Angervoiceclap_attribute:Anger0.975-0.030.765-0.450.9920.506
135Arousalvoiceclap_attribute:Arousal1.861+0.711.759+0.541.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.954+0.931.001+1.030.5090.479
137Authenticityvoiceclap_attribute:Authenticity3.304+3.193.273+3.012.7700.167
138Awevoiceclap_attribute:Awe0.605+0.250.747+0.800.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.241-1.240.332-0.890.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.411-0.770.278-1.330.5920.236
141Concentrationvoiceclap_attribute:Concentration1.259+0.081.246+0.061.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.498+0.921.463+0.841.0690.468
143Confusionvoiceclap_attribute:Confusion0.977+1.000.916+0.820.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.831+0.130.720-0.150.7800.406
145Contemptvoiceclap_attribute:Contempt0.578-0.380.372-0.990.7090.339
146Contentmentvoiceclap_attribute:Contentment0.526-0.200.641+0.170.5880.307
147Disappointmentvoiceclap_attribute:Disappointment1.036+0.950.932+0.590.7600.291
148Disgustvoiceclap_attribute:Disgust0.641+0.080.641+0.080.6250.207
149Distressvoiceclap_attribute:Distress1.163+0.671.097+0.490.9270.350
150Doubtvoiceclap_attribute:Doubt0.686+0.810.662+0.680.5350.188
151Elationvoiceclap_attribute:Elation0.375-0.410.603+0.160.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.481+0.550.506+0.630.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.771+0.400.637-0.100.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.389-0.280.410-0.200.4620.261
155Fearvoiceclap_attribute:Fear0.860+1.140.768+0.710.6180.212
156Gendervoiceclap_attribute:Gender1.246+1.391.497+1.64-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.902+0.910.823+0.620.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.545-0.791.557-0.761.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism1.014+0.781.199+1.250.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability1.218+0.311.052+0.001.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.704+0.290.778+0.510.6080.332
162Interestvoiceclap_attribute:Interest1.815+0.491.883+0.651.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness-0.023-1.130.146-0.770.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.940+1.701.018+2.020.5230.245
165Longingvoiceclap_attribute:Longing0.793+1.160.790+1.150.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.584-0.720.553-0.810.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive2.185+0.692.341+0.951.7750.594
168Painvoiceclap_attribute:Pain0.569+0.860.468+0.400.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.493-0.040.511+0.030.5030.301
170Pridevoiceclap_attribute:Pride0.569-0.600.570-0.600.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.851+1.642.840+1.612.3130.328
172Reliefvoiceclap_attribute:Relief1.081+0.141.052+0.071.0250.415
173Sadnessvoiceclap_attribute:Sadness0.996+0.780.956+0.650.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.353-0.790.786-0.030.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.315-0.760.483-0.360.6350.419
176Shamevoiceclap_attribute:Shame1.030+1.780.859+1.170.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.666+1.410.737+1.600.1600.360
178Sournessvoiceclap_attribute:Sourness0.837+0.450.584-0.420.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.320-0.850.253-1.030.6230.359
180Teasingvoiceclap_attribute:Teasing0.490-0.130.521-0.050.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.575+0.361.266-0.221.3840.529
182Triumphvoiceclap_attribute:Triumph0.353-1.010.508-0.580.7160.359
183Valencevoiceclap_attribute:Valence-0.735-1.15-0.324-0.50-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached1.445-0.351.333-0.551.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold0.129+0.490.467+1.33-0.0660.400
186durationvoiceclap_attribute:duration9.273-0.269.879-0.1610.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.612+0.743.493+0.503.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.823+1.134.843+1.274.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.837+0.492.844+0.502.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.907+0.231.901+0.151.8910.069
191talking speedvoiceclap_attribute:talking_speed16.312+1.9416.488+1.999.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
0-0.0085-0.0250-0.0510
10.12750.06410.0687
20.11000.10480.1268
3-0.1078-0.0927-0.0837
4-0.0733-0.0660-0.0776
5-0.1484-0.1262-0.1044
60.11440.10390.0107
70.21390.20680.2114
8-0.0204-0.0256-0.0155
90.00990.02510.0809
10-0.0498-0.0556-0.0127
11-0.0716-0.1121-0.0965
12-0.0864-0.0684-0.0830
13-0.0773-0.0456-0.0345
140.14670.11600.1226
150.02760.02400.0482
16-0.0943-0.1005-0.1022
17-0.0768-0.0752-0.0858
18-0.0733-0.0670-0.0509
19-0.0952-0.0815-0.0781
20-0.0336-0.0227-0.0647
21-0.0195-0.0268-0.0441
22-0.0821-0.0665-0.0681
23-0.1668-0.1728-0.0977
24-0.0136-0.03830.0092
25-0.0616-0.0762-0.0725
26-0.0535-0.0587-0.0622
27-0.0502-0.0323-0.0770
28-0.0873-0.0927-0.0909
290.12660.13590.1008
300.03730.0393-0.0114
31-0.0548-0.0512-0.0205
320.21570.19610.1807
33-0.0156-0.0095-0.0603
34-0.0637-0.0743-0.0740
350.15190.19320.1493
36-0.0681-0.0544-0.0887
37-0.0851-0.0609-0.0977
380.08770.06750.1287
39-0.0960-0.0913-0.0833
400.06020.12330.0958
41-0.0404-0.0641-0.0944
42-0.0733-0.0680-0.0833
430.02840.02710.0038
440.03080.02450.0109
45-0.0083-0.01820.0247
460.17810.16600.0888
47-0.0925-0.0903-0.0834
48-0.0812-0.0830-0.0806
49-0.0332-0.0191-0.0352
500.04430.04050.0141
510.04410.05990.0713
52-0.0908-0.0801-0.0938
53-0.01420.0081-0.0687
54-0.0882-0.0854-0.0923
55-0.0227-0.0342-0.0059
56-0.0738-0.0869-0.0783
57-0.0535-0.0825-0.0842
580.03320.03520.0294
590.06810.12520.1281
600.10480.10000.0417
610.15450.12810.1769
620.02670.06310.0691
63-0.02130.0210-0.0030
640.04930.03790.0335
65-0.0668-0.0728-0.0673
660.12830.09470.1437
67-0.0659-0.0815-0.0796
680.04450.05990.0924
690.0234-0.02230.0260
700.03100.02450.0199
710.29340.29710.1877
72-0.0543-0.1184-0.0541
730.04260.05680.0141
74-0.01500.02390.0571
750.21830.17180.1991
760.12920.10100.1379
77-0.0698-0.0874-0.0754
78-0.0738-0.0699-0.0734
79-0.0764-0.1039-0.0716
80-0.0378-0.03470.0030
810.15890.14080.2004
82-0.0712-0.0561-0.0790
830.17810.21160.2322
84-0.1030-0.0723-0.0984
850.03300.12810.0779
86-0.0423-0.1005-0.0255
87-0.1056-0.1126-0.0953
88-0.02240.00010.0363
89-0.0421-0.0485-0.0502
900.06070.0680-0.0039
910.01220.00840.0443
92-0.0310-0.0374-0.0112
93-0.0515-0.0517-0.0699
94-0.0308-0.03710.0087
95-0.0404-0.0757-0.0887
96-0.0768-0.0767-0.0212
970.13180.12620.2073
98-0.0720-0.0515-0.0420
990.08210.06160.0712
1000.15720.15730.1707
101-0.0223-0.0002-0.0798
102-0.00320.05560.0797
103-0.1096-0.1199-0.0871
1040.03120.02220.0431
105-0.0851-0.0864-0.0796
1060.03080.05120.0764
107-0.0576-0.0811-0.0769
108-0.0864-0.0976-0.0895
109-0.0576-0.0830-0.0630
1100.0299-0.0054-0.0244
1110.03270.00330.0377
112-0.0742-0.0811-0.0814
113-0.0231-0.0313-0.0402
1140.10300.05950.0161
1150.13270.13880.1740
1160.09560.07570.1097
117-0.02440.00560.0373
118-0.0825-0.0859-0.0804
1190.05890.0502-0.0629
1200.05220.08300.0965
1210.00250.03980.0940
122-0.01480.0220-0.0032
1230.0094-0.0299-0.0024
1240.04930.02520.0444
125-0.0939-0.0284-0.0823
126-0.1070-0.0937-0.0712
1270.11180.09510.0695
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
00.03200.0814-0.0303
1-0.0422-0.0273-0.0363
2-0.06550.0134-0.0461
3-0.0246-0.0368-0.0490
40.06150.09290.1310
50.03180.0087-0.0392
60.05830.09000.0266
70.08380.14490.2192
80.0444-0.01490.0133
90.03690.04930.0682
10-0.0633-0.0028-0.0394
11-0.0807-0.0105-0.0358
120.0451-0.03020.0671
130.13300.15350.1812
14-0.0479-0.0330-0.0326
150.14810.11600.1469
160.07280.03450.0247
170.00650.0232-0.0325
180.12350.07300.1375
19-0.1065-0.0592-0.0595
200.01640.02470.0461
21-0.0113-0.0228-0.0348
220.04030.04960.1013
23-0.0586-0.0756-0.0363
24-0.0150-0.0036-0.0320
25-0.00580.0164-0.0440
26-0.00600.0105-0.0414
27-0.0304-0.0089-0.0305
28-0.01170.0128-0.0406
29-0.0249-0.00200.1152
300.04350.02970.1844
310.14870.15470.1786
320.12730.07560.0751
330.02540.06720.0415
340.15950.15810.0498
350.02980.07910.0569
360.07530.09000.0481
37-0.0914-0.0459-0.0259
38-0.0570-0.0713-0.0521
39-0.0592-0.0912-0.0559
400.0056-0.0181-0.0204
410.01490.03320.0125
42-0.01310.0008-0.0177
430.01760.03840.0253
440.01390.09930.0548
45-0.0826-0.0479-0.0282
46-0.1090-0.0941-0.0551
47-0.0608-0.0785-0.0174
480.0266-0.02350.0252
49-0.00830.02470.0869
50-0.00870.00950.0249
51-0.00660.04130.0394
520.05800.04910.0027
53-0.0655-0.0511-0.0363
540.00470.02810.0280
550.01710.0001-0.0103
560.0144-0.0430-0.0203
570.07720.05110.0499
58-0.03780.0088-0.0290
590.01130.02960.0221
600.0129-0.0024-0.0200
61-0.0908-0.0390-0.0349
62-0.0084-0.0410-0.0330
630.06900.09870.1071
64-0.0457-0.0724-0.0418
650.09830.04240.0696
660.15130.12060.1475
67-0.1223-0.1685-0.0395
680.0235-0.0162-0.0282
690.11280.08020.0057
700.04570.05340.0953
71-0.0084-0.0130-0.0337
720.02850.0024-0.0416
730.08630.07960.0165
74-0.0901-0.0871-0.0405
750.13170.09180.1119
760.11340.13270.0521
77-0.06050.02780.0492
780.0348-0.0073-0.0090
79-0.0291-0.0439-0.0402
800.01160.01590.1488
810.01260.08660.0787
82-0.0432-0.0785-0.0360
83-0.0416-0.0147-0.0333
840.02840.0129-0.0030
850.00590.0126-0.0376
86-0.0109-0.0065-0.0358
87-0.0615-0.0395-0.0320
880.10970.06260.1161
890.03920.00510.1053
900.00430.0189-0.0017
910.0476-0.00430.0503
920.04660.03200.0122
930.04760.06670.0652
94-0.00550.0131-0.0493
950.22190.17660.1650
960.06240.10960.1506
97-0.02080.04160.1021
98-0.04250.02910.0914
99-0.01510.01510.0391
1000.07590.01080.0381
1010.0057-0.0004-0.0108
102-0.1097-0.1143-0.0448
1030.0429-0.0096-0.0379
104-0.0495-0.0848-0.0522
105-0.0266-0.03100.0124
106-0.0432-0.0039-0.0024
107-0.0266-0.02290.0509
108-0.0369-0.0074-0.0298
109-0.04130.0180-0.0385
110-0.0488-0.0531-0.0179
1110.14370.12470.1610
112-0.0067-0.01060.0334
113-0.1204-0.1581-0.0683
114-0.0681-0.0730-0.0559
115-0.0071-0.0326-0.0118
116-0.01140.02680.0204
1170.0199-0.01070.0527
1180.01070.04880.0922
1190.05770.08660.1028
1200.00430.0597-0.0343
1210.07250.00340.0380
122-0.0239-0.01340.0296
1230.02300.04210.1013
124-0.0026-0.0030-0.0217
1250.16390.10450.0720
1260.07030.03510.0605
127-0.0383-0.04500.0615
128-0.0731-0.0964-0.0499
129-0.0529-0.0545-0.0425
1300.08320.0531-0.0122
131-0.0696-0.0338-0.0351
1320.05740.01050.0780
1330.02350.0540-0.0318
1340.02930.0658-0.0384
135-0.0838-0.0652-0.0403
136-0.0725-0.0465-0.0429
1370.03720.03000.0735
138-0.0362-0.0092-0.0353
1390.0030-0.0070-0.0410
140-0.05860.00090.0119
1410.06680.05080.0808
142-0.0373-0.0146-0.0346
1430.04730.01130.0219
144-0.01020.0017-0.0244
145-0.0599-0.0525-0.0509
1460.05260.04130.0421
147-0.0649-0.0580-0.0356
148-0.01480.0159-0.0442
1490.12100.15580.0748
1500.03260.08480.0393
151-0.01280.04410.0712
152-0.0596-0.0118-0.0190
1530.02880.0231-0.0168
154-0.02820.01050.0187
1550.06960.06520.0488
156-0.0542-0.0505-0.0401
1570.16130.17430.0739
1580.0296-0.0294-0.0245
1590.10840.11540.1732
160-0.0517-0.0175-0.0197
1610.04290.05450.0176
162-0.0684-0.0326-0.0465
163-0.00190.0302-0.0278
1640.0006-0.0303-0.0352
1650.05510.09290.0839
166-0.01240.0291-0.0212
1670.01090.01340.0069
168-0.0217-0.0000-0.0145
1690.04350.00040.0114
170-0.0558-0.09290.0390
1710.0284-0.0103-0.0352
172-0.00500.03560.0217
1730.05610.08250.0454
1740.01500.02320.0574
175-0.0403-0.0343-0.0663
176-0.0258-0.0525-0.0341
177-0.0197-0.0355-0.0330
1780.12920.14200.1076
1790.0118-0.0015-0.0493
1800.00900.09750.0743
181-0.0422-0.0554-0.0571
182-0.0162-0.01980.0373
183-0.0882-0.0768-0.0792
1840.07190.0531-0.0407
1850.02690.01180.0737
186-0.0605-0.0548-0.0432
1870.00690.04670.1021
1880.00390.07330.0590
1890.02160.09410.0515
1900.04190.0276-0.0039
1910.01430.0170-0.0002
1920.07470.0773-0.0485
193-0.04380.0030-0.0060
194-0.0298-0.00610.0011
195-0.05200.0276-0.0381
196-0.01530.0136-0.0072
1970.01470.07850.0621
198-0.0785-0.0427-0.0387
199-0.1128-0.0900-0.0299
200-0.0277-0.0672-0.0463
2010.05960.06980.1054
2020.01700.07100.0005
2030.0107-0.0450-0.0500
2040.04920.09580.1074
2050.00460.0252-0.0144
2060.11090.14490.0600
207-0.00710.03520.1318
2080.0060-0.0198-0.0415
209-0.0753-0.05280.0170
210-0.0416-0.0756-0.0521
211-0.00400.02610.0120
212-0.0014-0.03100.0721
2130.16260.13680.0809
2140.00620.05400.0716
2150.08190.05770.0901
2160.15380.13220.0778
2170.09770.09640.0524
218-0.1128-0.0854-0.0465
2190.00410.03000.0331
220-0.0845-0.0531-0.0442
221-0.0728-0.0819-0.0348
222-0.0520-0.03980.0144
223-0.00640.04850.0803
2240.03260.05630.1141
225-0.0851-0.0365-0.0321
2260.02570.03620.0235
2270.0555-0.00240.1272
228-0.03880.0303-0.0232
2290.08260.10680.0566
230-0.0087-0.0121-0.0007
2310.02850.04500.0645
2320.03750.03650.0125
2330.08320.08600.0581
2340.08630.11140.0605
2350.09080.09520.0058
236-0.0511-0.0364-0.0191
237-0.0353-0.0248-0.0409
238-0.01260.03810.0198
239-0.0043-0.0001-0.0414
240-0.0596-0.0534-0.0408
2410.0247-0.03690.0545
242-0.0671-0.0756-0.0590
243-0.0174-0.0084-0.0500
2440.04850.08020.1055
2450.0065-0.04670.0249
246-0.00280.00890.0438
247-0.0056-0.0378-0.0337
248-0.1298-0.0929-0.0458
249-0.00980.0185-0.0597
Source: HUMANENESS_SMALL_VOICE_IDENTITY_R1_R2_2026-10-02_anime_000.html · SHA-256 b0c1ea42224790d9…
g11GermanAnime voice study

anime_000 · de · 32/50

Generated TTS prediction · Rank 2 qkvo · 11.3 s

Wir müssen aufhören, das für diese ganze Generation neuer Leute wie ein reines Glücksspiel zu sehen. Es ist absolut entscheidend, dass wir diese Wahrnehmung jetzt ändern.

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Hope Enthusiasm Optimism
z +3.31 / +3.10raw 2.64 / 2.47
Contentment
z +1.95 / +1.74raw 1.00 / 0.89
Elation
z +1.85 / +1.56raw 1.36 / 1.15
Pleasure Ecstasy
z +1.54 / +1.33raw 0.96 / 0.83
Concentration
z +1.18 / +1.51raw 1.51 / 1.76
Relief
z +1.17 / +1.41raw 0.83 / 0.99
Contemplation
z +1.36 / +1.40raw 0.90 / 0.92

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Narrator style (S_NARR)
z +1.84 / +2.30raw 3.96 / 4.64
Vocal register (REGS)
z +1.65 / +1.69raw 3.30 / 3.35
ASMR style (S_ASMR)
z +1.18 / +1.52raw 3.04 / 3.52
Whispered talking style (S_WHIS)
z +1.15 / +1.40raw 3.45 / 3.87
Monologue style (S_MONO)
z +1.25 / +1.39raw 4.43 / 4.61
Storytelling style (S_STRY)
z +0.97 / +1.29raw 3.81 / 4.25
Respiration (RESP)
z +1.13 / +1.19raw 3.87 / 3.93

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z -0.40 / -0.48raw 1.68 / 1.55
Vocal-burst blend (0–10)z -0.73 / -0.92raw 4.69 / 4.15
Content enjoyment (CE)z +0.98 / +1.18raw 5.80 / 6.00
Production quality (PQ)z +0.41 / +0.41raw 7.40 / 7.41
Speech signal quality, calibratedz +0.08 / +0.13raw 3.53 / 3.55
Background quality, calibratedz +0.33 / +0.51raw 3.69 / 3.86
Overall quality, calibratedz +0.40 / +0.55raw 3.08 / 3.17
P.808 MOS estimatez +0.55 / +0.11raw 3.69 / 3.48
score content enjoymentz +0.91 / +0.57raw 4.78 / 4.68
score overall qualityz +0.51 / +0.51raw 2.91 / 2.91
Background Noisez -0.87 / -0.44raw 0.58 / 0.69
Tempo (TEMP)z -0.09 / -0.20raw 2.01 / 1.88
Arousal (AROU)z -0.99 / -1.11raw 1.22 / 1.08
Valence (VALN)z -0.87 / -1.10raw 1.36 / 1.10
Recording quality (RCQL)z +0.72 / +0.91raw 3.39 / 3.56
Background noise (BKGN)z +0.57 / +0.58raw 3.14 / 3.14
talking speedz +0.47 / +0.64raw 10.80 / 11.45
Characters per second (CPS): 15.04 raw · z +1.60

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.7900.846
Identity · 250D0.6800.741

Vocal-burst regions

Tiny
Base
0 s11.3 s

Verified burst intervals are unavailable for this clip.

Tiny: 0 regions, count-head estimate 0.02; Base: 0 regions, count-head estimate 0.01.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.520+0.900.798+1.390.0070.570
1Amusementemo_Amusement0.200+0.230.088+0.10-0.0010.885
2Angeremo_Anger0.110+0.170.176+0.280.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.198+0.340.081+0.130.0050.572
4Aweemo_Awe0.249+0.580.043+0.100.0000.427
5Bitternessemo_Bitterness0.018+0.01-0.003-0.040.0130.415
6Concentrationemo_Concentration1.510+1.181.755+1.510.6340.741
7Confusionemo_Confusion0.034+0.060.009+0.020.0010.540
8Contemplationemo_Contemplation0.897+1.360.920+1.400.0820.598
9Contemptemo_Contempt-0.064-0.14-0.034-0.07-0.0000.464
10Contentmentemo_Contentment1.004+1.950.893+1.740.0000.514
11Disappointmentemo_Disappointment0.299+0.430.536+0.770.0000.700
12Disgustemo_Disgust0.027+0.040.011+0.010.0070.489
13Distressemo_Distress0.304+0.370.550+0.670.0000.821
14Doubtemo_Doubt0.040+0.070.046+0.090.0060.466
15Elationemo_Elation1.361+1.851.151+1.560.0120.728
16Embarrassmentemo_Embarrassment-0.037-0.19-0.017-0.140.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.098+0.100.126+0.130.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion-0.011-0.070.024-0.010.0280.539
19Fearemo_Fear0.345+0.540.637+1.020.0150.610
20Helplessnessemo_Helplessness0.107+0.160.322+0.47-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism2.637+3.312.471+3.100.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.175+0.040.335+0.260.1490.705
23Infatuationemo_Infatuation0.008+0.040.027+0.08-0.0150.533
24Interestemo_Interest2.585+1.102.413+0.911.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness0.013+0.07-0.015+0.02-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.110+0.120.072+0.030.0590.442
27Longingemo_Longing0.390+0.480.557+0.740.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.071+0.030.067+0.020.0550.534
29Painemo_Pain0.013+0.03-0.000+0.00-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.962+1.540.831+1.330.0170.614
31Prideemo_Pride0.389+0.760.387+0.760.0020.509
32Reliefemo_Relief0.828+1.170.992+1.410.0330.681
33Sadnessemo_Sadness0.499+0.660.859+1.130.0000.759
34Sexual Lustemo_Sexual_Lust0.021-0.020.056+0.050.0300.494
35Shameemo_Shame-0.001-0.030.002-0.020.0130.510
36Sournessemo_Sourness0.032+0.08-0.023-0.04-0.0050.439
37Teasingemo_Teasing-0.035-0.06-0.070-0.130.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.359+0.460.543+0.720.0410.697
39Triumphemo_Triumph0.529+1.030.513+1.00-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg3.814+0.674.040+0.852.9841.238
41Arousal (AROU)vn_AROU_reg1.222-0.991.081-1.112.3591.147
42Arousal shift (ARSH)vn_ARSH_reg1.136-1.661.068-1.742.5800.867
43Vocal attack (ATCK)vn_ATCK_reg0.941-0.940.679-1.152.1621.295
44Background noise (BKGN)vn_BKGN_reg3.137+0.573.142+0.582.7030.761
45Brightness (BRGT)vn_BRGT_reg1.865-0.151.725-0.331.9910.813
46Speech chunking (CHNK)vn_CHNK_reg1.843-0.001.921+0.061.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg2.155+0.152.039+0.022.0180.945
48Cognitive load (COGL)vn_COGL_reg3.104+0.952.890+0.752.0941.061
49Dynamic arc (DARC)vn_DARC_reg0.336-0.930.187-1.041.6151.380
50Disfluency (DFLU)vn_DFLU_reg1.779-0.601.583-0.762.4881.184
51Emphasis (EMPH)vn_EMPH_reg2.831+0.152.865+0.172.6071.524
52Esthetics (ESTH)vn_ESTH_reg3.116+0.913.238+1.072.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.109+0.070.170+0.260.0860.322
54Focus (FOCS)vn_FOCS_reg2.287-0.052.504+0.152.3441.102
55Fullness (FULL)vn_FULL_reg2.262+0.542.105+0.371.7780.889
56Perceived gender (GEND)vn_GEND_reg1.119-1.851.020-1.914.0551.588
57Harmonicity (HARM)vn_HARM_reg1.185-1.551.165-1.572.4800.836
58Metallic character (METL)vn_METL_reg0.210-1.470.016-1.681.6210.958
59Pitch range (RANG)vn_RANG_reg3.295+0.433.344+0.482.9140.897
60Recording quality (RCQL)vn_RCQL_reg3.392+0.723.560+0.912.7540.888
61Vocal register (REGS)vn_REGS_reg3.297+1.653.346+1.691.1981.268
62Respiration (RESP)vn_RESP_reg3.874+1.133.933+1.192.6801.056
63Roughness (ROUG)vn_ROUG_reg1.575-0.521.468-0.642.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg1.298-0.991.146-1.172.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg2.162+0.022.277+0.162.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg1.891-0.021.730-0.181.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg1.180-0.301.036-0.471.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg2.049-0.712.000-0.762.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg1.687-0.671.484-0.942.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg1.398-0.841.389-0.852.0700.803
71Smoothness (SMTH)vn_SMTH_reg2.519+0.342.801+0.602.1431.093
72Stance (STNC)vn_STNC_reg1.466-1.031.627-0.882.6331.137
73Speech structure (STRU)vn_STRU_reg2.141+0.002.197+0.062.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg3.041+1.183.515+1.521.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg0.897-0.180.891-0.191.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg1.326-0.171.337-0.161.5601.401
77Casual style (S_CASU)vn_S_CASU_reg3.506-0.223.314-0.373.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg1.891-0.531.779-0.622.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg3.389+0.583.620+0.782.7211.148
80Formal style (S_FORM)vn_S_FORM_reg2.010+0.801.927+0.721.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg4.425+1.254.614+1.392.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg3.964+1.844.637+2.301.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.453+0.080.568+0.280.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg1.872-0.361.613-0.562.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg1.611+0.311.571+0.281.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg3.812+0.974.247+1.292.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg0.825+0.150.749+0.080.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg3.453+1.153.865+1.401.5971.618
89Tempo (TEMP)vn_TEMP_reg2.013-0.091.876-0.202.1171.190
90Tension (TENS)vn_TENS_reg1.221-0.401.016-0.641.5590.852
91Valence (VALN)vn_VALN_reg1.356-0.871.096-1.102.3281.118
92Valence shift (VALS)vn_VALS_reg1.528-0.921.575-0.872.5181.079
93Velocity flux (VFLX)vn_VFLX_reg1.189-0.881.137-0.932.2561.208
94Volatility (VOLT)vn_VOLT_reg3.214+0.712.991+0.462.5860.886
95Vulnerability (VULN)vn_VULN_reg3.746+1.123.491+0.862.6480.984
96Warmth (WARM)vn_WARM_reg3.079+0.553.217+0.702.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_61.683-0.401.551-0.482.3321.621
98Vocal-burst blend (0–10)blend_0_104.690-0.734.149-0.926.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.756-1.000.684-1.221.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.050-0.683.243-0.023.2500.293
101Arousaleiv_extra:Arousal2.029+1.702.073+1.801.2890.435
102Authenticityeiv_extra:Authenticity3.000+2.183.030+2.372.6670.153
103Background Noiseeiv_extra:Background_Noise0.578-0.870.690-0.440.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.131+0.301.095+0.181.0400.303
105Gendereiv_extra:Gender1.224+1.851.294+1.92-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.682-0.581.618-0.841.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive2.616+1.612.499+1.361.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.689+1.642.699+1.672.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous1.061+0.390.816-0.330.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.743+1.440.825+1.700.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.595+0.680.687+0.960.3760.324
112Valenceeiv_extra:Valence0.596+1.550.515+1.340.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.487-0.521.364-0.801.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold0.716+2.220.735+2.280.0510.300
115score background qualityeiv_extra:score_background_quality3.515+0.313.461+0.233.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.783+0.914.679+0.574.5070.303
117score overall qualityeiv_extra:score_overall_quality2.909+0.512.911+0.512.6520.507
118score speech qualityeiv_extra:score_speech_quality1.796-0.081.748-0.411.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.800+0.985.999+1.184.7991.018
120Content usefulness (CU)audiobox:CU6.358+0.716.737+1.095.6491.001
121Production complexity (PC)audiobox:PC1.877-0.282.016-0.192.3311.617
122Production quality (PQ)audiobox:PQ7.400+0.417.407+0.417.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw3.967+0.104.007+0.163.9000.669
124Background quality, rawdnsmos:BAK_raw3.735+0.473.973+0.683.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.459+0.463.595+0.633.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.528+0.083.553+0.133.4880.501
127Background quality, calibrateddnsmos:BAK3.694+0.333.865+0.513.3890.928
128Overall quality, calibrateddnsmos:OVRL3.084+0.403.172+0.552.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.692+0.553.475+0.113.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.018+0.030.010+0.020.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection1.266+0.571.083+0.250.9470.554
132Agevoiceclap_attribute:Age3.203-0.053.269+0.073.2290.547
133Amusementvoiceclap_attribute:Amusement0.324-0.120.236-0.270.3920.572
134Angervoiceclap_attribute:Anger0.869-0.240.789-0.400.9920.506
135Arousalvoiceclap_attribute:Arousal1.487+0.111.431+0.021.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.818+0.650.806+0.620.5090.479
137Authenticityvoiceclap_attribute:Authenticity3.216+2.673.220+2.692.7700.167
138Awevoiceclap_attribute:Awe0.647+0.420.699+0.620.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.419-0.560.465-0.390.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.596+0.010.560-0.140.5920.236
141Concentrationvoiceclap_attribute:Concentration1.182-0.061.347+0.251.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.312+0.521.387+0.681.0690.468
143Confusionvoiceclap_attribute:Confusion1.116+1.421.156+1.530.6430.334
144Contemplationvoiceclap_attribute:Contemplation1.164+0.951.215+1.070.7800.406
145Contemptvoiceclap_attribute:Contempt0.452-0.760.399-0.920.7090.339
146Contentmentvoiceclap_attribute:Contentment0.728+0.460.611+0.070.5880.307
147Disappointmentvoiceclap_attribute:Disappointment1.059+1.030.989+0.790.7600.291
148Disgustvoiceclap_attribute:Disgust0.686+0.300.669+0.220.6250.207
149Distressvoiceclap_attribute:Distress1.481+1.591.402+1.360.9270.350
150Doubtvoiceclap_attribute:Doubt0.617+0.430.622+0.460.5350.188
151Elationvoiceclap_attribute:Elation0.537-0.000.623+0.210.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.425+0.380.423+0.370.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.808+0.540.739+0.280.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.557+0.370.601+0.530.4620.261
155Fearvoiceclap_attribute:Fear1.091+2.231.120+2.370.6180.212
156Gendervoiceclap_attribute:Gender1.301+1.441.513+1.66-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness1.164+1.881.086+1.600.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.440-1.071.426-1.111.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.874+0.430.887+0.460.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability0.764-0.530.731-0.591.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.881+0.820.842+0.700.6080.332
162Interestvoiceclap_attribute:Interest1.731+0.301.818+0.501.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.513+0.020.296-0.440.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.969+1.820.965+1.810.5230.245
165Longingvoiceclap_attribute:Longing0.859+1.460.879+1.550.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.631-0.580.653-0.520.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive1.971+0.331.980+0.341.7750.594
168Painvoiceclap_attribute:Pain0.759+1.730.742+1.650.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.526+0.080.494-0.030.5030.301
170Pridevoiceclap_attribute:Pride0.588-0.550.550-0.660.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.679+1.122.725+1.262.3130.328
172Reliefvoiceclap_attribute:Relief1.875+2.051.586+1.351.0250.415
173Sadnessvoiceclap_attribute:Sadness1.193+1.411.115+1.160.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.679-0.220.587-0.380.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.554-0.190.717+0.200.6350.419
176Shamevoiceclap_attribute:Shame1.123+2.111.145+2.190.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.610+1.250.574+1.150.1600.360
178Sournessvoiceclap_attribute:Sourness0.553-0.530.464-0.830.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.160-1.290.183-1.230.6230.359
180Teasingvoiceclap_attribute:Teasing0.387-0.400.304-0.620.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.629+0.461.579+0.371.3840.529
182Triumphvoiceclap_attribute:Triumph0.781+0.180.682-0.090.7160.359
183Valencevoiceclap_attribute:Valence-0.069-0.10-0.044-0.06-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached1.379-0.471.447-0.341.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold0.146+0.530.082+0.37-0.0660.400
186durationvoiceclap_attribute:duration11.199+0.0410.578-0.0510.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.474+0.463.362+0.243.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.753+0.664.739+0.564.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.841+0.502.824+0.462.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.875-0.231.850-0.591.8910.069
191talking speedvoiceclap_attribute:talking_speed10.801+0.4711.446+0.649.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.0831-0.0076-0.0588
10.12330.12250.0989
20.07380.14250.1511
3-0.0671-0.0783-0.0825
4-0.0764-0.0677-0.0765
5-0.1055-0.1155-0.1030
60.11890.05560.0494
70.26330.29360.2554
80.01630.0144-0.0577
90.00070.00380.0381
10-0.0559-0.07180.0187
11-0.0543-0.0885-0.0952
12-0.0952-0.1090-0.0818
13-0.1016-0.0804-0.0594
140.03130.05130.0925
150.05780.06770.0388
16-0.0415-0.0950-0.0761
17-0.1163-0.0874-0.0846
18-0.0518-0.0939-0.0166
19-0.0914-0.0783-0.0770
200.07090.0613-0.0205
210.06420.0173-0.0034
22-0.0188-0.0389-0.0136
23-0.0572-0.0934-0.0963
24-0.0336-0.01030.0137
25-0.0543-0.0424-0.0715
26-0.0655-0.0712-0.0763
27-0.0374-0.0683-0.0759
28-0.0984-0.1106-0.0896
290.12400.14030.0810
300.04860.0367-0.0382
31-0.0786-0.0691-0.0228
320.17900.20080.1400
330.08950.0386-0.0255
34-0.0658-0.0594-0.0729
350.17380.15000.1022
36-0.1138-0.0842-0.0875
37-0.0032-0.0119-0.0963
380.00630.02520.1288
39-0.0305-0.0240-0.0820
40-0.01620.05100.0477
410.03640.0125-0.0501
42-0.0863-0.0874-0.0821
43-0.05530.0079-0.0107
440.13040.0362-0.0077
45-0.0006-0.00660.0428
460.17380.16620.0689
47-0.0812-0.0842-0.0823
48-0.0946-0.0864-0.0795
490.0010-0.0629-0.0364
500.09710.08690.0224
510.00360.07880.0983
52-0.0169-0.0537-0.0667
530.01620.0318-0.0525
54-0.0959-0.1020-0.0911
550.0024-0.05400.0078
56-0.0674-0.0847-0.0771
57-0.0674-0.0356-0.0815
58-0.0447-0.05720.0646
590.03930.04480.1351
600.12460.14570.0638
610.19430.23100.1831
620.00790.05260.0392
630.0038-0.0251-0.0517
640.07250.07180.0304
65-0.0837-0.0793-0.0663
660.0454-0.03270.1222
67-0.0748-0.0891-0.0784
680.05880.07230.0456
690.06900.06040.1361
700.02960.01180.0084
710.24670.25370.2203
72-0.0524-0.05180.0130
730.05050.02310.0092
74-0.0751-0.01440.0575
750.24290.15440.2014
760.08440.08310.1512
77-0.0585-0.0837-0.0743
78-0.0761-0.0702-0.0724
79-0.00940.0313-0.0348
80-0.0483-0.0491-0.0580
810.25310.23530.2664
82-0.0388-0.0277-0.0779
830.06230.10630.1764
84-0.0844-0.1025-0.0804
850.01550.07660.0552
860.1214-0.05370.0643
87-0.1029-0.1446-0.0939
88-0.0559-0.0353-0.0331
89-0.0677-0.0604-0.0746
900.04280.02970.0368
910.0091-0.00080.0334
92-0.0876-0.0316-0.0352
93-0.0169-0.0564-0.0351
940.03130.0124-0.0168
95-0.0291-0.0982-0.0874
96-0.0978-0.0766-0.0697
970.07960.09710.1711
98-0.0645-0.0389-0.0682
990.09910.08960.1105
1000.04920.02940.1428
101-0.0296-0.0212-0.0756
102-0.01790.02890.0634
103-0.1048-0.1036-0.0859
1040.09840.05640.0128
105-0.0550-0.0977-0.0784
106-0.01190.01190.0775
107-0.0626-0.0847-0.0758
108-0.0684-0.0917-0.0883
109-0.0171-0.0645-0.0332
1100.0096-0.0192-0.0354
1110.00320.00360.1302
112-0.0377-0.0478-0.0802
113-0.0620-0.0907-0.0684
1140.13870.14570.1082
1150.18410.18890.1560
1160.06170.04970.0927
117-0.0267-0.01370.0291
118-0.0751-0.0777-0.0792
1190.12210.0772-0.0239
1200.09910.11980.1127
121-0.00660.00150.0535
122-0.0168-0.0486-0.0711
123-0.0585-0.0529-0.0320
1240.08370.03510.0701
125-0.0863-0.0783-0.0811
126-0.0537-0.0583-0.0701
1270.16040.07930.0350
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.0523-0.0197-0.0278
1-0.0446-0.0334-0.0386
2-0.0729-0.0625-0.0491
3-0.0446-0.0198-0.0387
40.05710.06480.1344
50.00490.0193-0.0417
60.09450.14230.0683
70.03830.05850.1210
80.11250.06420.0305
90.02130.03180.0522
10-0.0691-0.0394-0.0419
11-0.0708-0.0244-0.0540
120.04100.07050.0523
130.06290.11640.2114
14-0.0228-0.0200-0.0272
150.10040.11570.1958
160.01490.0725-0.0195
170.00000.0170-0.0479
180.05790.04560.1665
19-0.0800-0.0984-0.0633
200.0034-0.0195-0.0087
21-0.0115-0.0264-0.0370
220.01980.05520.0511
23-0.0750-0.0998-0.0386
240.0337-0.0106-0.0341
250.0233-0.0043-0.0201
260.05830.0582-0.0440
27-0.0132-0.0371-0.0324
280.02000.0161-0.0432
29-0.0348-0.02810.0399
30-0.0385-0.01740.0116
31-0.03870.06950.0959
320.16490.11970.1041
330.06460.07350.0144
340.17910.20090.1536
350.06540.01680.0249
360.07540.04690.0511
37-0.0991-0.1151-0.0276
38-0.0554-0.0016-0.0477
390.0073-0.0456-0.0594
400.02560.01040.0358
41-0.01070.0264-0.0275
420.11410.10110.0688
430.03170.02560.0007
440.0003-0.0039-0.0268
45-0.1133-0.1111-0.0385
46-0.0741-0.1104-0.0586
47-0.0414-0.0765-0.0472
480.0229-0.02230.0884
490.0176-0.00770.0425
50-0.00680.03290.0496
51-0.03500.02580.0607
520.04370.0411-0.0220
53-0.0523-0.0384-0.0386
540.07410.06950.0325
55-0.00300.0314-0.0323
56-0.0203-0.01550.0565
570.04580.09440.0876
58-0.0491-0.0103-0.0383
590.07080.06680.0706
600.03120.04360.0085
61-0.0385-0.0346-0.0370
620.01310.0495-0.0337
63-0.0113-0.01210.0601
64-0.0845-0.0891-0.0444
650.06290.03940.1215
660.07540.11240.1377
670.0228-0.0210-0.0420
68-0.0429-0.0313-0.0124
690.04520.08350.1010
700.05000.08350.0359
71-0.0253-0.0078-0.0418
72-0.0675-0.0565-0.0442
730.10160.11970.0641
74-0.0708-0.0898-0.0431
750.05310.08050.1130
760.10830.11240.1051
77-0.0646-0.0439-0.0102
780.05830.06350.0106
79-0.0650-0.0885-0.0427
80-0.0650-0.0595-0.0170
81-0.01080.0095-0.0422
820.0023-0.0288-0.0182
83-0.0410-0.0542-0.0354
840.02130.03330.0306
850.03000.03240.0150
860.05000.0140-0.0381
870.0068-0.0209-0.0340
880.07330.11040.1746
890.04580.06120.0628
900.06120.03860.0468
910.0145-0.00820.1179
92-0.0059-0.0009-0.0190
93-0.0571-0.0042-0.0484
94-0.0410-0.0555-0.0690
950.13160.16630.1619
96-0.01520.00980.1174
97-0.0883-0.02690.0300
98-0.04460.05750.0734
99-0.03790.01090.0418
1000.17740.13040.1234
1010.12580.07050.0259
102-0.0331-0.0589-0.0426
1030.10580.08210.0563
104-0.0354-0.0781-0.0555
105-0.0637-0.0898-0.0554
106-0.0537-0.0185-0.0328
1070.03810.02290.0043
1080.03580.0224-0.0317
109-0.04270.0130-0.0410
110-0.0331-0.0519-0.0507
1110.23160.19150.2098
112-0.0030-0.01100.0170
1130.0008-0.1111-0.0442
114-0.0741-0.0274-0.0594
115-0.0766-0.0404-0.0302
1160.02980.03840.0336
1170.03670.03560.0556
1180.0155-0.0308-0.0191
1190.10000.10770.1247
120-0.02710.0120-0.0543
1210.06080.00710.0128
122-0.0469-0.0251-0.0402
1230.00240.02410.0349
1240.06250.04290.0569
1250.15990.14230.1061
1260.10500.06750.0719
127-0.0483-0.0449-0.0267
128-0.0298-0.0248-0.0274
129-0.0354-0.0645-0.0452
1300.06910.08010.0346
131-0.0479-0.0539-0.0373
1320.11000.06090.1318
1330.02850.0449-0.0421
1340.09330.0569-0.0243
135-0.0354-0.0402-0.0369
136-0.0099-0.0321-0.0457
1370.03870.0354-0.0199
138-0.0416-0.0619-0.0376
1390.04020.0256-0.0436
1400.02620.06620.0114
1410.12740.0752-0.0063
142-0.0441-0.0429-0.0368
143-0.0072-0.0397-0.0370
1440.00020.0243-0.0289
145-0.0900-0.0998-0.0541
146-0.0140-0.02490.0280
147-0.0762-0.0752-0.0378
1480.01030.0288-0.0470
1490.07210.08450.0856
150-0.0167-0.0099-0.0149
151-0.0406-0.03670.0100
152-0.0554-0.0271-0.0472
1530.01210.0269-0.0062
1540.06870.03820.0388
1550.06330.08910.1648
1560.0190-0.0032-0.0427
1570.13080.19690.0620
1580.09620.04110.0341
1590.02810.01750.0446
160-0.0479-0.04260.0128
161-0.01500.0264-0.0307
162-0.1054-0.0865-0.0494
1630.0177-0.0170-0.0474
1640.02550.0010-0.0374
165-0.01580.05190.0826
1660.01130.0150-0.0258
1670.08580.0223-0.0054
168-0.0317-0.02000.0394
1690.08160.09780.0511
170-0.0285-0.0542-0.0399
1710.02480.0108-0.0160
172-0.0245-0.0136-0.0213
1730.12410.10510.0858
174-0.0400-0.0082-0.0359
175-0.0491-0.0755-0.0701
1760.0024-0.0166-0.0362
177-0.0431-0.0412-0.0351
1780.01330.07450.0275
179-0.01860.0211-0.0525
1800.0044-0.0479-0.0119
1810.00370.0329-0.0235
182-0.0224-0.0269-0.0368
183-0.0128-0.0294-0.0859
1840.01040.0022-0.0433
185-0.01440.03390.0093
186-0.0508-0.0442-0.0514
187-0.0733-0.03280.0685
1880.00390.03620.0472
1890.02620.04520.0033
1900.08620.0555-0.0007
1910.0020-0.0076-0.0184
192-0.0051-0.0010-0.0083
193-0.0829-0.0432-0.0449
194-0.0596-0.0251-0.0552
195-0.0654-0.0111-0.0405
1960.03620.05020.0536
197-0.00760.0532-0.0080
198-0.0381-0.0662-0.0412
199-0.0937-0.0991-0.0468
200-0.0138-0.0379-0.0492
2010.05330.09640.1035
202-0.01420.0436-0.0122
203-0.0129-0.0394-0.0531
204-0.03690.02860.0515
2050.06460.06920.0837
2060.13080.11440.1085
207-0.00860.01450.0365
208-0.0171-0.0185-0.0442
209-0.0477-0.03430.0174
210-0.0120-0.0414-0.0554
211-0.0398-0.00920.0220
212-0.0259-0.03080.0132
2130.18080.15760.1536
2140.03810.03710.0718
2150.06370.01530.0337
2160.24320.19690.1761
2170.05540.11370.0634
218-0.1391-0.1357-0.0494
219-0.0502-0.0050-0.0554
220-0.0558-0.0492-0.0470
2210.0224-0.0115-0.0365
222-0.0448-0.0688-0.0326
223-0.05410.00100.0523
224-0.00690.00790.0714
225-0.0360-0.04590.0425
2260.07910.05590.0495
2270.0273-0.01010.0993
228-0.0163-0.0582-0.0689
2290.07460.05890.0447
230-0.03310.0074-0.0434
2310.05710.10040.0010
2320.02580.03620.0048
233-0.0246-0.04290.0494
2340.03120.0841-0.0310
2350.00110.04290.0256
236-0.0296-0.0284-0.0366
237-0.0239-0.0192-0.0435
238-0.00560.00280.0160
239-0.0175-0.0485-0.0440
240-0.0820-0.0429-0.0434
2410.09200.08080.1407
242-0.0641-0.0602-0.0627
243-0.0712-0.0645-0.0547
2440.05020.07680.0565
2450.01540.02960.0955
246-0.0261-0.0210-0.0241
2470.0077-0.0191-0.0359
248-0.1020-0.1024-0.0487
249-0.0162-0.0376-0.0033
Source: HUMANENESS_SMALL_VOICE_IDENTITY_R1_R2_2026-10-02_anime_000.html · SHA-256 f177f287bee08aa4…
g12GermanAnime voice study

anime_000 · de · 50/50

Generated TTS prediction · Rank 1 qv · 14.3 s

Ein perfekt trainierter Jagdsaker im Traum zu besitzen, deutet darauf hin, dass du Macht übernimmst oder schlimmer noch, zu einem Tyrannen wirst. Es spricht von einer schleichenden Ungerechtigkeit, einer spirituellen Blindheit, der du dich stellen musst.

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Contemplation
z +1.60 / +1.90raw 1.04 / 1.22
Concentration
z +1.37 / +1.75raw 1.65 / 1.93
Hope Enthusiasm Optimism
z +1.54 / +1.41raw 1.29 / 1.19
Contentment
z +1.37 / +0.77raw 0.70 / 0.40
Awe
z +1.29 / +1.14raw 0.55 / 0.49
Interest
z +1.07 / +1.23raw 2.56 / 2.71
Fear
z +0.50 / +1.12raw 0.32 / 0.70

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Narrator style (S_NARR)
z +2.66 / +2.88raw 5.15 / 5.47
Vocal register (REGS)
z +2.27 / +2.34raw 4.07 / 4.17
Storytelling style (S_STRY)
z +1.94 / +1.83raw 5.13 / 4.99
Whispered talking style (S_WHIS)
z +1.60 / +1.55raw 4.18 / 4.10
Dramatic style (S_DRAM)
z +1.42 / +1.37raw 4.35 / 4.29
ASMR style (S_ASMR)
z +1.13 / +1.32raw 2.97 / 3.23
Pitch range (RANG)
z +1.19 / +1.29raw 3.98 / 4.07

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z -0.68 / -0.68raw 1.23 / 1.23
Vocal-burst blend (0–10)z -0.77 / -0.83raw 4.56 / 4.41
Content enjoyment (CE)z +0.95 / +1.34raw 5.77 / 6.17
Production quality (PQ)z +0.27 / +0.33raw 7.28 / 7.34
Speech signal quality, calibratedz +0.03 / +0.07raw 3.50 / 3.53
Background quality, calibratedz +0.40 / +0.58raw 3.76 / 3.92
Overall quality, calibratedz +0.41 / +0.57raw 3.09 / 3.19
P.808 MOS estimatez +0.47 / +0.36raw 3.66 / 3.60
score content enjoymentz +1.08 / +0.97raw 4.84 / 4.80
score overall qualityz +0.55 / +0.53raw 2.93 / 2.92
Background Noisez -0.95 / -0.99raw 0.56 / 0.55
Tempo (TEMP)z +0.76 / +0.79raw 3.02 / 3.05
Arousal (AROU)z -0.25 / -0.39raw 2.07 / 1.91
Valence (VALN)z -0.43 / -0.59raw 1.85 / 1.67
Recording quality (RCQL)z +0.93 / +1.04raw 3.58 / 3.68
Background noise (BKGN)z +0.76 / +0.80raw 3.28 / 3.31
talking speedz +0.55 / +0.74raw 11.10 / 11.81
Characters per second (CPS): 17.79 raw · z +2.33

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8330.861
Identity · 250D0.6890.774

Vocal-burst regions

Tiny
Base
0 s14.3 s

Verified burst intervals are unavailable for this clip.

Tiny: 0 regions, count-head estimate 0.03; Base: 0 regions, count-head estimate 0.12.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.564+0.980.325+0.560.0070.570
1Amusementemo_Amusement0.149+0.170.007+0.01-0.0010.885
2Angeremo_Anger0.134+0.210.166+0.260.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.277+0.480.140+0.230.0050.572
4Aweemo_Awe0.551+1.290.487+1.140.0000.427
5Bitternessemo_Bitterness-0.013-0.06-0.011-0.060.0130.415
6Concentrationemo_Concentration1.647+1.371.932+1.750.6340.741
7Confusionemo_Confusion0.109+0.200.138+0.250.0010.540
8Contemplationemo_Contemplation1.042+1.601.219+1.900.0820.598
9Contemptemo_Contempt0.037+0.080.033+0.07-0.0000.464
10Contentmentemo_Contentment0.705+1.370.398+0.770.0000.514
11Disappointmentemo_Disappointment0.109+0.160.180+0.260.0000.700
12Disgustemo_Disgust0.033+0.050.021+0.030.0070.489
13Distressemo_Distress0.050+0.060.111+0.140.0000.821
14Doubtemo_Doubt0.185+0.390.448+0.950.0060.466
15Elationemo_Elation0.370+0.490.166+0.210.0120.728
16Embarrassmentemo_Embarrassment-0.037-0.19-0.032-0.170.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.225+0.240.109+0.110.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.010-0.030.027-0.000.0280.539
19Fearemo_Fear0.322+0.500.699+1.120.0150.610
20Helplessnessemo_Helplessness0.005+0.010.112+0.17-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism1.292+1.541.188+1.410.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.028-0.170.182+0.050.1490.705
23Infatuationemo_Infatuation0.073+0.160.001+0.03-0.0150.533
24Interestemo_Interest2.556+1.072.706+1.231.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness-0.001+0.04-0.011+0.02-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.103+0.100.056-0.010.0590.442
27Longingemo_Longing0.287+0.320.226+0.220.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.176+0.230.203+0.280.0550.534
29Painemo_Pain-0.028-0.05-0.019-0.03-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.336+0.520.128+0.180.0170.614
31Prideemo_Pride0.112+0.220.104+0.200.0020.509
32Reliefemo_Relief0.084+0.070.032-0.000.0330.681
33Sadnessemo_Sadness0.196+0.260.262+0.350.0000.759
34Sexual Lustemo_Sexual_Lust0.142+0.230.132+0.210.0300.494
35Shameemo_Shame-0.024-0.07-0.013-0.050.0130.510
36Sournessemo_Sourness0.342+0.790.183+0.43-0.0050.439
37Teasingemo_Teasing0.461+0.850.367+0.670.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.176+0.190.068+0.040.0410.697
39Triumphemo_Triumph0.034+0.080.016+0.04-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg3.567+0.473.622+0.512.9841.238
41Arousal (AROU)vn_AROU_reg2.071-0.251.909-0.392.3591.147
42Arousal shift (ARSH)vn_ARSH_reg1.852-0.841.568-1.172.5800.867
43Vocal attack (ATCK)vn_ATCK_reg1.481-0.531.633-0.412.1621.295
44Background noise (BKGN)vn_BKGN_reg3.281+0.763.311+0.802.7030.761
45Brightness (BRGT)vn_BRGT_reg2.263+0.332.279+0.351.9910.813
46Speech chunking (CHNK)vn_CHNK_reg2.408+0.452.598+0.611.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg2.754+0.782.803+0.832.0180.945
48Cognitive load (COGL)vn_COGL_reg2.402+0.292.483+0.372.0941.061
49Dynamic arc (DARC)vn_DARC_reg1.541-0.051.331-0.211.6151.380
50Disfluency (DFLU)vn_DFLU_reg1.320-0.991.159-1.122.4881.184
51Emphasis (EMPH)vn_EMPH_reg3.264+0.433.338+0.482.6071.524
52Esthetics (ESTH)vn_ESTH_reg3.231+1.063.347+1.212.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.156+0.220.090+0.010.0860.322
54Focus (FOCS)vn_FOCS_reg3.173+0.752.978+0.582.3441.102
55Fullness (FULL)vn_FULL_reg1.799+0.021.859+0.091.7780.889
56Perceived gender (GEND)vn_GEND_reg0.734-2.090.858-2.014.0551.588
57Harmonicity (HARM)vn_HARM_reg1.172-1.571.172-1.572.4800.836
58Metallic character (METL)vn_METL_reg0.629-1.040.517-1.151.6210.958
59Pitch range (RANG)vn_RANG_reg3.978+1.194.069+1.292.9140.897
60Recording quality (RCQL)vn_RCQL_reg3.578+0.933.680+1.042.7540.888
61Vocal register (REGS)vn_REGS_reg4.072+2.274.170+2.341.1981.268
62Respiration (RESP)vn_RESP_reg3.569+0.843.406+0.692.6801.056
63Roughness (ROUG)vn_ROUG_reg1.454-0.651.262-0.872.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg1.060-1.271.166-1.152.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg3.070+1.123.115+1.182.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg2.239+0.332.149+0.241.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg0.781-0.780.765-0.801.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg3.196+0.392.969+0.172.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg1.603-0.781.484-0.942.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg1.464-0.761.691-0.472.0700.803
71Smoothness (SMTH)vn_SMTH_reg2.922+0.713.026+0.812.1431.093
72Stance (STNC)vn_STNC_reg1.778-0.751.791-0.742.6331.137
73Speech structure (STRU)vn_STRU_reg2.692+0.582.715+0.602.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg2.971+1.133.235+1.321.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg1.059-0.031.089+0.001.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg2.765+0.862.573+0.721.5601.401
77Casual style (S_CASU)vn_S_CASU_reg3.167-0.493.003-0.623.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg2.419-0.132.429-0.132.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg4.355+1.424.292+1.372.7211.148
80Formal style (S_FORM)vn_S_FORM_reg2.427+1.212.443+1.221.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg4.087+0.984.217+1.082.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg5.153+2.665.467+2.881.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.550+0.250.743+0.590.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg2.802+0.342.501+0.112.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg2.167+0.741.776+0.441.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg5.129+1.944.988+1.832.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg1.097+0.370.773+0.100.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg4.177+1.604.102+1.551.5971.618
89Tempo (TEMP)vn_TEMP_reg3.025+0.763.053+0.792.1171.190
90Tension (TENS)vn_TENS_reg1.742+0.211.574+0.021.5590.852
91Valence (VALN)vn_VALN_reg1.851-0.431.665-0.592.3281.118
92Valence shift (VALS)vn_VALS_reg1.955-0.522.054-0.432.5181.079
93Velocity flux (VFLX)vn_VFLX_reg2.029-0.191.825-0.362.2561.208
94Volatility (VOLT)vn_VOLT_reg3.410+0.933.271+0.772.5860.886
95Vulnerability (VULN)vn_VULN_reg3.587+0.953.723+1.092.6480.984
96Warmth (WARM)vn_WARM_reg2.691+0.122.593+0.022.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_61.231-0.681.231-0.682.3321.621
98Vocal-burst blend (0–10)blend_0_104.563-0.774.414-0.836.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.673-1.250.655-1.301.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.185-0.223.255+0.013.2500.293
101Arousaleiv_extra:Arousal1.678+0.891.755+1.071.2890.435
102Authenticityeiv_extra:Authenticity2.992+2.132.998+2.172.6670.153
103Background Noiseeiv_extra:Background_Noise0.556-0.950.545-0.990.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.046+0.021.161+0.401.0400.303
105Gendereiv_extra:Gender1.263+1.881.185+1.81-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.759-0.271.725-0.411.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive2.308+0.952.232+0.791.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.699+1.672.728+1.762.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous0.843-0.250.689-0.710.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.612+1.030.542+0.820.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.473+0.300.510+0.420.3760.324
112Valenceeiv_extra:Valence0.166+0.420.260+0.670.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.671-0.101.618-0.221.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold0.260+0.700.220+0.560.0510.300
115score background qualityeiv_extra:score_background_quality3.478+0.253.496+0.283.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.836+1.084.800+0.974.5070.303
117score overall qualityeiv_extra:score_overall_quality2.931+0.552.919+0.532.6520.507
118score speech qualityeiv_extra:score_speech_quality1.765-0.291.761-0.321.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.768+0.956.166+1.344.7991.018
120Content usefulness (CU)audiobox:CU6.136+0.496.475+0.835.6491.001
121Production complexity (PC)audiobox:PC1.852-0.301.833-0.312.3311.617
122Production quality (PQ)audiobox:PQ7.279+0.277.335+0.337.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw3.929+0.043.977+0.123.9000.669
124Background quality, rawdnsmos:BAK_raw3.837+0.564.034+0.733.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.478+0.483.630+0.673.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.505+0.033.525+0.073.4880.501
127Background quality, calibrateddnsmos:BAK3.760+0.403.923+0.583.3890.928
128Overall quality, calibrateddnsmos:OVRL3.089+0.413.188+0.572.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.655+0.473.602+0.363.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.030+0.050.109+0.170.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection1.063+0.210.936-0.020.9470.554
132Agevoiceclap_attribute:Age3.230+0.003.396+0.313.2290.547
133Amusementvoiceclap_attribute:Amusement0.418+0.050.308-0.150.3920.572
134Angervoiceclap_attribute:Anger1.019+0.050.818-0.350.9920.506
135Arousalvoiceclap_attribute:Arousal1.747+0.531.748+0.531.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.911+0.841.018+1.060.5090.479
137Authenticityvoiceclap_attribute:Authenticity3.245+2.843.319+3.292.7700.167
138Awevoiceclap_attribute:Awe0.669+0.500.833+1.130.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.424-0.540.425-0.540.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.544-0.210.519-0.310.5920.236
141Concentrationvoiceclap_attribute:Concentration1.391+0.331.477+0.501.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.069+0.001.296+0.481.0690.468
143Confusionvoiceclap_attribute:Confusion0.805+0.480.860+0.650.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.863+0.201.074+0.730.7800.406
145Contemptvoiceclap_attribute:Contempt0.592-0.350.408-0.890.7090.339
146Contentmentvoiceclap_attribute:Contentment0.675+0.280.654+0.210.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.968+0.721.017+0.880.7600.291
148Disgustvoiceclap_attribute:Disgust0.692+0.330.647+0.110.6250.207
149Distressvoiceclap_attribute:Distress1.186+0.741.320+1.120.9270.350
150Doubtvoiceclap_attribute:Doubt0.575+0.210.631+0.510.5350.188
151Elationvoiceclap_attribute:Elation0.600+0.150.640+0.250.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.282-0.080.344+0.120.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.686+0.080.566-0.370.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.364-0.370.486+0.090.4620.261
155Fearvoiceclap_attribute:Fear1.009+1.841.186+2.680.6180.212
156Gendervoiceclap_attribute:Gender1.458+1.601.623+1.76-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness1.036+1.411.099+1.640.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.650-0.511.614-0.601.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism1.021+0.800.956+0.630.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability0.882-0.310.697-0.661.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.697+0.270.729+0.370.6080.332
162Interestvoiceclap_attribute:Interest2.069+1.072.117+1.181.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.182-0.690.078-0.910.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.983+1.881.020+2.030.5230.245
165Longingvoiceclap_attribute:Longing0.735+0.900.877+1.540.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.781-0.150.583-0.720.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive2.086+0.522.123+0.591.7750.594
168Painvoiceclap_attribute:Pain0.650+1.230.750+1.690.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.505+0.010.513+0.030.5030.301
170Pridevoiceclap_attribute:Pride0.702-0.220.595-0.530.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.902+1.802.912+1.832.3130.328
172Reliefvoiceclap_attribute:Relief1.583+1.351.586+1.351.0250.415
173Sadnessvoiceclap_attribute:Sadness0.926+0.561.006+0.810.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.422-0.670.385-0.740.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.490-0.350.444-0.460.6350.419
176Shamevoiceclap_attribute:Shame0.934+1.441.072+1.930.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.533+1.040.571+1.140.1600.360
178Sournessvoiceclap_attribute:Sourness0.777+0.250.645-0.210.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.394-0.640.185-1.220.6230.359
180Teasingvoiceclap_attribute:Teasing0.577+0.100.399-0.370.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.459+0.141.331-0.101.3840.529
182Triumphvoiceclap_attribute:Triumph0.675-0.110.668-0.130.7160.359
183Valencevoiceclap_attribute:Valence-0.097-0.14-0.001+0.01-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached1.514-0.221.329-0.561.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold-0.013+0.130.094+0.40-0.0660.400
186durationvoiceclap_attribute:duration13.466+0.4014.802+0.6110.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.534+0.583.519+0.553.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.882+1.544.866+1.434.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.847+0.512.895+0.632.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.855-0.521.841-0.711.8910.069
191talking speedvoiceclap_attribute:talking_speed11.101+0.5511.812+0.749.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.05300.0218-0.0332
10.09240.10120.0636
20.10330.17190.1589
3-0.0705-0.0800-0.0786
4-0.0776-0.0686-0.0729
5-0.1394-0.1175-0.0981
60.14100.0544-0.0193
70.24920.29170.2291
8-0.0307-0.0683-0.0636
90.05000.05060.1155
10-0.0721-0.07510.0196
11-0.0582-0.0623-0.0907
12-0.0973-0.0963-0.0780
130.0037-0.0014-0.0387
140.10170.08110.1641
15-0.00790.07560.0213
16-0.0552-0.0805-0.0961
17-0.0924-0.0631-0.0806
18-0.0776-0.0920-0.0308
19-0.0973-0.0838-0.0734
20-0.0112-0.0035-0.0776
210.03070.0215-0.0208
22-0.0265-0.0672-0.0729
23-0.0831-0.0843-0.0917
240.0098-0.00490.0708
25-0.0516-0.0482-0.0681
26-0.0675-0.0822-0.0728
27-0.0399-0.0724-0.0723
28-0.0858-0.1061-0.0854
290.09350.13550.0883
300.02380.0318-0.0351
31-0.0924-0.0702-0.0303
320.21100.30690.2078
330.0413-0.0143-0.0780
34-0.0853-0.0713-0.0695
350.14430.15340.0721
36-0.0913-0.0963-0.0834
37-0.0519-0.0427-0.0919
380.07320.06010.1635
39-0.0612-0.0465-0.0782
400.05470.13660.0624
410.0148-0.0156-0.0667
42-0.0956-0.0990-0.0782
43-0.05360.0170-0.0115
440.07100.03510.0094
450.0005-0.02600.0310
460.14320.17740.0809
47-0.1006-0.0936-0.0784
48-0.0989-0.0930-0.0757
490.0228-0.0490-0.0476
500.09890.08870.0278
510.00560.05550.0831
52-0.0522-0.0952-0.0881
53-0.0434-0.0392-0.0791
54-0.1022-0.0985-0.0868
550.0068-0.0724-0.0136
56-0.0864-0.0974-0.0735
57-0.0492-0.0800-0.0804
580.0008-0.00660.1025
590.06070.10120.0822
600.04920.07890.0392
610.13500.13880.1570
620.00020.04630.0803
630.07820.0438-0.0024
640.04840.04220.0058
65-0.0749-0.0784-0.0633
660.12020.04430.1741
67-0.0393-0.0658-0.0748
680.09240.11370.0712
690.06890.04650.0835
700.03660.0033-0.0019
710.27540.23830.1801
72-0.0174-0.02840.0124
730.06860.01970.0377
74-0.05410.00490.0735
750.28640.23400.1991
760.12730.09140.1700
77-0.0721-0.1045-0.0709
78-0.0754-0.0650-0.0690
79-0.0072-0.0212-0.0673
80-0.0721-0.0501-0.0066
810.21640.13770.1962
82-0.0710-0.0686-0.0742
830.13880.13280.2120
84-0.0147-0.0642-0.0962
850.02420.04760.0183
860.0896-0.02920.0255
87-0.0199-0.0865-0.0696
88-0.0495-0.01820.0177
89-0.0967-0.0547-0.0712
900.06230.02830.0185
910.0150-0.0226-0.0047
92-0.0568-0.0087-0.0041
93-0.0443-0.0525-0.0385
940.0440-0.0231-0.0074
95-0.0128-0.0735-0.0754
96-0.0609-0.0729-0.0615
970.15630.14470.1787
98-0.0287-0.0403-0.0289
990.12840.06560.1306
1000.13830.07730.1915
101-0.0283-0.0212-0.0750
102-0.02400.02760.0391
103-0.0743-0.0985-0.0818
1040.09780.0729-0.0008
105-0.0675-0.0833-0.0748
1060.00020.02860.0533
107-0.0598-0.0996-0.0207
108-0.1066-0.0974-0.0842
109-0.0038-0.0446-0.0300
1100.05140.0386-0.0156
1110.03010.03250.0983
112-0.0329-0.0555-0.0765
113-0.0269-0.0729-0.0464
1140.06640.04650.0253
1150.17270.17090.1705
1160.11370.04410.0965
1170.00630.00720.0312
118-0.0754-0.1001-0.0755
1190.0284-0.0223-0.0750
1200.08580.11150.0989
1210.07760.02420.0979
122-0.01870.0099-0.0434
1230.01110.03160.0173
1240.02790.02290.0771
125-0.0902-0.0683-0.0773
126-0.0579-0.0838-0.0668
1270.13010.06040.0191
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.0557-0.0252-0.0417
1-0.0691-0.0294-0.0367
2-0.0628-0.0238-0.0466
3-0.0485-0.0558-0.0594
40.02420.07680.1180
50.07200.0871-0.0396
60.15380.09790.0649
7-0.01990.03810.1372
80.08260.01290.0047
9-0.00760.00920.0628
10-0.07750.0206-0.0398
11-0.0468-0.0100-0.0383
120.07230.04100.0373
130.14400.19570.2305
14-0.0430-0.0535-0.0541
150.17330.18090.2202
160.09010.0629-0.0236
17-0.04220.0299-0.0455
180.05280.04720.1243
19-0.0697-0.0922-0.0602
200.00820.04040.0306
21-0.0237-0.0404-0.0352
220.05770.05040.0889
23-0.0663-0.0762-0.0367
24-0.0231-0.0040-0.0324
25-0.0714-0.0052-0.0271
260.01950.0606-0.0418
27-0.03130.0198-0.0308
28-0.0107-0.0063-0.0410
29-0.0505-0.02970.0467
30-0.01300.00460.0660
31-0.02710.04640.1030
320.18360.14570.0922
33-0.00530.0543-0.0340
340.17670.20710.1143
350.02710.02570.0381
360.00920.0277-0.0091
37-0.0519-0.0248-0.0262
38-0.0405-0.0279-0.0490
39-0.0471-0.0256-0.0565
40-0.00600.01810.0226
41-0.02830.05520.0198
420.07570.04890.0789
430.07520.05580.0315
44-0.02040.0043-0.0457
45-0.0683-0.0560-0.0362
46-0.0866-0.1058-0.0557
470.0337-0.0242-0.0479
480.03560.02420.0676
49-0.0186-0.03870.0234
500.05800.08250.0458
510.02880.05920.0548
52-0.03370.00170.0010
53-0.0390-0.0211-0.0367
540.04560.03840.0552
550.05220.0159-0.0545
56-0.00050.00340.0745
570.05970.04550.0667
58-0.04500.0119-0.0364
590.02190.03430.0726
600.06940.06060.0414
61-0.0290-0.0364-0.0352
62-0.00760.0734-0.0333
630.01130.01530.0582
64-0.0792-0.0643-0.0422
650.08320.05920.0820
660.06080.12350.1089
67-0.0217-0.0780-0.0399
68-0.01050.0074-0.0275
690.06660.12180.1050
700.01220.04130.0713
71-0.0217-0.0250-0.0397
72-0.0786-0.0566-0.0420
730.09350.06290.0428
74-0.0674-0.0859-0.0409
750.06770.05890.1045
760.10040.12350.1038
770.01010.00910.0337
780.02700.0100-0.0208
79-0.0534-0.0734-0.0406
80-0.0387-0.03560.0040
810.01530.0512-0.0148
820.0089-0.0226-0.0364
83-0.0545-0.0053-0.0337
840.00480.03290.0347
850.02750.04500.0341
860.00690.0163-0.0362
87-0.0565-0.0213-0.0323
880.03430.06340.1255
890.02480.04070.0742
900.09240.0142-0.0219
910.00880.03340.0696
92-0.0582-0.0092-0.0350
93-0.02190.0185-0.0212
94-0.03790.0216-0.0314
950.15840.22870.1869
960.02020.05720.0963
97-0.05970.04350.1025
980.01600.12060.1429
990.05540.04440.0766
1000.13080.10410.1364
1010.07980.06710.0342
102-0.0614-0.0814-0.0452
1030.11530.03340.0392
104-0.0448-0.0586-0.0528
105-0.0689-0.0669-0.0527
106-0.0110-0.0165-0.0312
1070.0077-0.0024-0.0247
1080.0110-0.0084-0.0301
109-0.04820.0367-0.0389
110-0.0280-0.0447-0.0482
1110.15950.20370.1934
112-0.01080.01920.0255
113-0.0763-0.1349-0.0690
114-0.0689-0.0164-0.0565
115-0.0792-0.0045-0.0287
1160.05050.02520.0395
1170.03360.02300.0498
1180.01760.00230.0080
1190.06140.08710.0740
120-0.03500.0229-0.0516
1210.02340.02450.0266
122-0.0363-0.0180-0.0382
1230.03840.02890.0682
1240.07060.0213-0.0212
1250.14170.14050.1068
1260.11710.05630.0557
127-0.0502-0.0632-0.0291
128-0.0384-0.0413-0.0244
129-0.0164-0.0802-0.0429
1300.08610.12180.0638
131-0.0436-0.0498-0.0354
1320.09010.05060.1261
1330.08210.10190.0125
1340.04220.0444-0.0202
135-0.05310.0032-0.0407
136-0.0207-0.0404-0.0434
137-0.00760.04320.0068
138-0.0430-0.0518-0.0357
139-0.0146-0.0044-0.0414
1400.02140.06710.0554
1410.06540.06570.0390
142-0.0663-0.0334-0.0350
1430.00180.0069-0.0065
144-0.04680.0038-0.0422
145-0.0763-0.0646-0.0515
1460.01470.03160.0461
147-0.0691-0.0620-0.0360
1480.00210.0049-0.0446
1490.15610.12120.0786
150-0.0029-0.0015-0.0399
151-0.0373-0.00640.0203
152-0.0476-0.0427-0.0449
1530.07290.02960.0041
1540.00310.0370-0.0169
1550.16640.13090.1390
156-0.0445-0.0220-0.0406
1570.09240.15590.0764
1580.0360-0.0349-0.0267
159-0.0393-0.0119-0.0107
160-0.0711-0.0216-0.0181
161-0.00060.0535-0.0234
162-0.0585-0.0313-0.0464
163-0.0597-0.0084-0.0451
164-0.0208-0.0090-0.0356
165-0.03840.02400.0927
1660.00670.0063-0.0294
1670.01410.0364-0.0307
1680.00640.02600.0756
1690.10270.06030.0165
170-0.0732-0.0569-0.0246
1710.03310.05060.0027
172-0.0330-0.0098-0.0320
1730.07400.07230.0842
174-0.0482-0.03700.0395
175-0.0640-0.0643-0.0671
176-0.0093-0.0452-0.0344
177-0.0666-0.0350-0.0334
1780.00420.12180.0072
179-0.04650.0266-0.0499
1800.00320.01080.0001
1810.02320.0025-0.0001
182-0.0516-0.0415-0.0350
183-0.0468-0.0637-0.0817
184-0.00620.0277-0.0411
1850.02510.01420.0508
186-0.0221-0.0179-0.0149
1870.00290.04750.1191
1880.02410.02250.0202
189-0.00890.03140.0074
1900.05740.0438-0.0078
1910.0051-0.0189-0.0541
192-0.0402-0.0173-0.0477
193-0.0709-0.0161-0.0077
194-0.0700-0.0270-0.0307
195-0.0832-0.0115-0.0385
1960.03290.0376-0.0158
1970.03470.09270.0428
198-0.0416-0.0814-0.0391
199-0.0551-0.0945-0.0445
200-0.0855-0.0415-0.0087
2010.07110.09100.1005
2020.04530.06340.0111
2030.0337-0.0114-0.0505
204-0.00160.08020.0954
2050.01460.04470.0835
2060.10840.11550.0657
207-0.04250.01560.0414
208-0.0390-0.0073-0.0420
209-0.0640-0.0435-0.0078
210-0.0798-0.0615-0.0527
211-0.01310.00620.0365
212-0.0161-0.0452-0.0015
2130.22840.17750.1530
214-0.00370.02390.0811
215-0.01130.01710.0621
2160.15380.14570.1265
2170.10330.08360.0891
218-0.1274-0.0495-0.0470
219-0.0740-0.0152-0.0527
2200.0096-0.0136-0.0447
2210.00850.0026-0.0250
222-0.0344-0.0283-0.0242
223-0.03670.05800.0853
224-0.01700.01400.0621
2250.0393-0.00060.0282
2260.08890.09050.1033
227-0.0184-0.02550.0914
228-0.0442-0.0336-0.0655
2290.02880.0538-0.0055
230-0.0680-0.0067-0.0412
2310.04020.06570.0155
232-0.0031-0.0326-0.0245
2330.11930.06660.0511
2340.09300.0899-0.0102
2350.04590.06090.0372
236-0.0290-0.0307-0.0348
237-0.0184-0.0266-0.0413
2380.01950.03600.0696
239-0.0562-0.0694-0.0418
240-0.0792-0.0344-0.0413
2410.03900.08190.1503
242-0.0775-0.0802-0.0596
243-0.0471-0.0728-0.0520
2440.14690.18660.1290
245-0.0014-0.02940.1075
246-0.0230-0.0188-0.0328
2470.0159-0.0292-0.0341
248-0.0519-0.0575-0.0463
249-0.0050-0.0095-0.0210
Source: HUMANENESS_SMALL_VOICE_IDENTITY_R1_R2_2026-10-02_anime_000.html · SHA-256 0bd8b4976ce0ed25…
e01GermanEmolia / Emilia

Emolia DE · DE_B00000_S00000_W000042

Natural speech · Original speech · 11.7 s

Und man arbeitet an Rabenkrieg. Das heißt, ich nehme an, die Zähne des Kamans werden dann der Anfang sein. Oder man baut vielleicht direkt die ganze Kampagne in einem nach, aber ich kann's mir fast nicht vorstellen.

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Contemplation
z +1.72 / +1.80raw 1.11 / 1.16
Concentration
z +1.60 / +1.53raw 1.82 / 1.77
Doubt
z +1.11 / +1.43raw 0.52 / 0.67
Fatigue Exhaustion
z +0.46 / +0.97raw 0.28 / 0.55
Impatience and Irritability
z +0.72 / +0.89raw 0.66 / 0.78
Anger
z +0.76 / +0.86raw 0.47 / 0.53
Emotional Numbness
z +0.66 / +0.60raw 0.61 / 0.55

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Mixed resonance (R_MIXD)
z +2.23 / +2.37raw 3.31 / 3.43
Formal style (S_FORM)
z +1.57 / +1.72raw 2.80 / 2.95
Authoritative style (S_AUTH)
z +1.36 / +1.50raw 2.50 / 2.66
Teacher or didactic style (S_TECH)
z +1.20 / +1.48raw 2.12 / 2.46
Newsreader style (S_NEWS)
z +1.18 / +1.47raw 1.08 / 1.24
Throat resonance (R_THRT)
z +1.38 / +1.35raw 3.18 / 3.16
Fullness (FULL)
z +1.30 / +1.34raw 2.93 / 2.97

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z +0.60 / +0.54raw 3.31 / 3.21
Vocal-burst blend (0–10)z -2.21 / -2.24raw 0.63 / 0.55
Content enjoyment (CE)z +0.26 / +0.54raw 5.07 / 5.35
Production quality (PQ)z -0.18 / +0.01raw 6.88 / 7.05
Speech signal quality, calibratedz +0.13 / +0.03raw 3.55 / 3.50
Background quality, calibratedz +0.45 / +0.47raw 3.80 / 3.83
Overall quality, calibratedz +0.48 / +0.52raw 3.13 / 3.16
P.808 MOS estimatez +0.78 / +0.96raw 3.81 / 3.90
score content enjoymentz +0.61 / +0.77raw 4.69 / 4.74
score overall qualityz +0.58 / +0.65raw 2.95 / 2.98
Background Noisez -0.50 / -0.84raw 0.68 / 0.58
Tempo (TEMP)z +0.06 / +0.01raw 2.19 / 2.12
Arousal (AROU)z +0.17 / +0.12raw 2.55 / 2.50
Valence (VALN)z +0.38 / +0.39raw 2.75 / 2.76
Recording quality (RCQL)z +0.25 / +0.13raw 2.97 / 2.87
Background noise (BKGN)z +0.35 / +0.48raw 2.97 / 3.07
talking speedz +2.31 / +1.96raw 17.69 / 16.37
Characters per second (CPS): 18.41 raw · z +2.50

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8960.900
Identity · 250D0.8240.844

Vocal-burst regions

Tiny
Base
0 s11.7 s

Verified burst intervals are unavailable for this clip.

Tiny: 1 regions, count-head estimate 1.08; Base: 1 regions, count-head estimate 0.97.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
Tiny14.945.280.34
Base14.945.260.32
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection-0.003-0.020.051+0.080.0070.570
1Amusementemo_Amusement0.090+0.100.064+0.07-0.0010.885
2Angeremo_Anger0.467+0.760.531+0.860.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.031+0.04-0.013-0.030.0050.572
4Aweemo_Awe0.023+0.05-0.004-0.010.0000.427
5Bitternessemo_Bitterness0.185+0.410.220+0.500.0130.415
6Concentrationemo_Concentration1.824+1.601.772+1.530.6340.741
7Confusionemo_Confusion0.336+0.620.219+0.400.0010.540
8Contemplationemo_Contemplation1.108+1.721.159+1.800.0820.598
9Contemptemo_Contempt0.080+0.170.077+0.17-0.0000.464
10Contentmentemo_Contentment0.241+0.470.192+0.370.0000.514
11Disappointmentemo_Disappointment0.214+0.310.424+0.610.0000.700
12Disgustemo_Disgust0.177+0.350.188+0.370.0070.489
13Distressemo_Distress0.098+0.120.186+0.230.0000.821
14Doubtemo_Doubt0.523+1.110.671+1.430.0060.466
15Elationemo_Elation0.000-0.020.037+0.030.0120.728
16Embarrassmentemo_Embarrassment-0.013-0.13-0.032-0.170.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.612+0.660.552+0.600.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.277+0.460.550+0.970.0280.539
19Fearemo_Fear0.131+0.190.146+0.220.0150.610
20Helplessnessemo_Helplessness0.121+0.180.235+0.34-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.039-0.100.080-0.050.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.657+0.720.779+0.890.1490.705
23Infatuationemo_Infatuation-0.035-0.04-0.021-0.01-0.0150.533
24Interestemo_Interest2.124+0.592.063+0.521.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness-0.052-0.05-0.035-0.02-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.091+0.070.064+0.010.0590.442
27Longingemo_Longing0.147+0.100.344+0.410.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.064+0.020.051-0.010.0550.534
29Painemo_Pain0.010+0.020.012+0.02-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.020+0.010.004-0.020.0170.614
31Prideemo_Pride-0.012-0.030.001-0.000.0020.509
32Reliefemo_Relief0.026-0.010.036+0.000.0330.681
33Sadnessemo_Sadness0.107+0.140.271+0.360.0000.759
34Sexual Lustemo_Sexual_Lust-0.019-0.10-0.037-0.130.0300.494
35Shameemo_Shame0.002-0.020.000-0.020.0130.510
36Sournessemo_Sourness0.199+0.460.206+0.48-0.0050.439
37Teasingemo_Teasing0.020+0.04-0.005-0.010.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude-0.015-0.080.031-0.010.0410.697
39Triumphemo_Triumph-0.036-0.06-0.034-0.05-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg3.289+0.253.346+0.292.9841.238
41Arousal (AROU)vn_AROU_reg2.554+0.172.499+0.122.3591.147
42Arousal shift (ARSH)vn_ARSH_reg2.613+0.042.535-0.052.5800.867
43Vocal attack (ATCK)vn_ATCK_reg2.125-0.032.204+0.032.1621.295
44Background noise (BKGN)vn_BKGN_reg2.971+0.353.066+0.482.7030.761
45Brightness (BRGT)vn_BRGT_reg2.622+0.782.496+0.621.9910.813
46Speech chunking (CHNK)vn_CHNK_reg3.060+0.982.992+0.921.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg1.952-0.072.198+0.192.0180.945
48Cognitive load (COGL)vn_COGL_reg2.277+0.172.343+0.232.0941.061
49Dynamic arc (DARC)vn_DARC_reg1.741+0.091.732+0.081.6151.380
50Disfluency (DFLU)vn_DFLU_reg3.309+0.693.249+0.642.4881.184
51Emphasis (EMPH)vn_EMPH_reg2.997+0.263.055+0.292.6071.524
52Esthetics (ESTH)vn_ESTH_reg2.824+0.542.988+0.752.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.123+0.120.115+0.090.0860.322
54Focus (FOCS)vn_FOCS_reg2.006-0.312.011-0.302.3441.102
55Fullness (FULL)vn_FULL_reg2.931+1.302.973+1.341.7780.889
56Perceived gender (GEND)vn_GEND_reg3.501-0.353.406-0.414.0551.588
57Harmonicity (HARM)vn_HARM_reg2.722+0.292.758+0.332.4800.836
58Metallic character (METL)vn_METL_reg1.490-0.141.508-0.121.6210.958
59Pitch range (RANG)vn_RANG_reg2.380-0.602.279-0.712.9140.897
60Recording quality (RCQL)vn_RCQL_reg2.973+0.252.866+0.132.7540.888
61Vocal register (REGS)vn_REGS_reg0.588-0.480.544-0.521.1981.268
62Respiration (RESP)vn_RESP_reg2.044-0.602.019-0.632.6801.056
63Roughness (ROUG)vn_ROUG_reg1.627-0.461.558-0.542.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg2.157+0.032.178+0.052.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg2.436+0.352.415+0.322.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg2.387+0.482.423+0.511.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg3.312+2.233.429+2.371.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg2.146-0.622.103-0.662.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg2.963+1.052.957+1.042.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg3.182+1.383.156+1.352.0700.803
71Smoothness (SMTH)vn_SMTH_reg3.209+0.983.200+0.972.1431.093
72Stance (STNC)vn_STNC_reg2.278-0.312.440-0.172.6331.137
73Speech structure (STRU)vn_STRU_reg2.892+0.793.023+0.922.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg1.337-0.021.318-0.041.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg2.502+1.362.656+1.501.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg-0.802-1.69-0.759-1.661.5601.401
77Casual style (S_CASU)vn_S_CASU_reg3.337-0.363.304-0.383.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg2.714+0.092.570-0.022.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg1.587-0.991.604-0.972.7211.148
80Formal style (S_FORM)vn_S_FORM_reg2.797+1.572.947+1.721.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg3.968+0.893.894+0.832.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg2.854+1.072.877+1.091.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg1.077+1.181.241+1.470.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg0.816-1.170.572-1.352.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg-0.938-1.67-1.197-1.871.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg1.718-0.561.569-0.672.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg2.118+1.202.460+1.480.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg1.578-0.011.802+0.131.5971.618
89Tempo (TEMP)vn_TEMP_reg2.191+0.062.125+0.012.1171.190
90Tension (TENS)vn_TENS_reg1.051-0.601.182-0.441.5590.852
91Valence (VALN)vn_VALN_reg2.751+0.382.764+0.392.3281.118
92Valence shift (VALS)vn_VALS_reg2.873+0.332.809+0.272.5181.079
93Velocity flux (VFLX)vn_VFLX_reg1.547-0.591.632-0.522.2561.208
94Volatility (VOLT)vn_VOLT_reg1.266-1.491.192-1.572.5860.886
95Vulnerability (VULN)vn_VULN_reg2.338-0.322.307-0.352.6480.984
96Warmth (WARM)vn_WARM_reg2.264-0.352.213-0.402.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_63.309+0.603.211+0.542.3321.621
98Vocal-burst blend (0–10)blend_0_100.635-2.210.550-2.246.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.627-1.390.593-1.491.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.375+0.423.383+0.453.2500.293
101Arousaleiv_extra:Arousal1.033-0.591.108-0.421.2890.435
102Authenticityeiv_extra:Authenticity2.727+0.392.802+0.882.6670.153
103Background Noiseeiv_extra:Background_Noise0.676-0.500.584-0.840.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.311+0.901.184+0.481.0400.303
105Gendereiv_extra:Gender-1.044-0.42-1.017-0.40-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched2.118+1.162.225+1.591.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive1.824-0.071.715-0.311.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.186+0.132.294+0.452.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous0.966+0.110.956+0.080.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.091-0.590.125-0.490.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.387+0.040.507+0.410.3760.324
112Valenceeiv_extra:Valence-0.009-0.040.027+0.060.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached2.089+0.862.208+1.131.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold-0.031-0.27-0.056-0.360.0510.300
115score background qualityeiv_extra:score_background_quality3.561+0.383.616+0.453.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.691+0.614.741+0.774.5070.303
117score overall qualityeiv_extra:score_overall_quality2.947+0.582.982+0.652.6520.507
118score speech qualityeiv_extra:score_speech_quality1.847+0.271.874+0.461.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.068+0.265.347+0.544.7991.018
120Content usefulness (CU)audiobox:CU5.694+0.055.654+0.015.6491.001
121Production complexity (PC)audiobox:PC1.770-0.351.845-0.302.3311.617
122Production quality (PQ)audiobox:PQ6.883-0.187.054+0.017.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw4.025+0.193.960+0.093.9000.669
124Background quality, rawdnsmos:BAK_raw3.907+0.623.973+0.683.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.538+0.563.599+0.633.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.555+0.133.502+0.033.4880.501
127Background quality, calibrateddnsmos:BAK3.803+0.453.825+0.473.3890.928
128Overall quality, calibrateddnsmos:OVRL3.131+0.483.155+0.522.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.810+0.783.899+0.963.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.732+1.130.676+1.040.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.414-0.960.344-1.090.9470.554
132Agevoiceclap_attribute:Age3.241+0.023.396+0.313.2290.547
133Amusementvoiceclap_attribute:Amusement0.086-0.54-0.069-0.810.3920.572
134Angervoiceclap_attribute:Anger0.678-0.620.525-0.920.9920.506
135Arousalvoiceclap_attribute:Arousal1.078-0.550.866-0.891.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise-0.041-1.150.056-0.940.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.682-0.532.741-0.182.7700.167
138Awevoiceclap_attribute:Awe-0.008-2.110.168-1.430.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.431-0.520.284-1.070.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.365-0.970.382-0.900.5920.236
141Concentrationvoiceclap_attribute:Concentration2.076+1.622.093+1.651.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.471+0.861.125+0.121.0690.468
143Confusionvoiceclap_attribute:Confusion0.688+0.130.532-0.330.6430.334
144Contemplationvoiceclap_attribute:Contemplation1.532+1.851.510+1.800.7800.406
145Contemptvoiceclap_attribute:Contempt0.375-0.990.353-1.050.7090.339
146Contentmentvoiceclap_attribute:Contentment0.318-0.880.278-1.010.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.454-1.050.563-0.680.7600.291
148Disgustvoiceclap_attribute:Disgust0.345-1.350.322-1.460.6250.207
149Distressvoiceclap_attribute:Distress0.620-0.880.319-1.740.9270.350
150Doubtvoiceclap_attribute:Doubt0.635+0.530.629+0.500.5350.188
151Elationvoiceclap_attribute:Elation0.237-0.760.078-1.160.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.078-0.730.023-0.900.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.897+0.870.773+0.410.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.582+0.460.501+0.150.4620.261
155Fearvoiceclap_attribute:Fear0.596-0.100.339-1.320.6180.212
156Gendervoiceclap_attribute:Gender-1.038-0.88-1.171-1.01-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.598-0.220.345-1.170.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched2.138+0.792.278+1.171.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.364-0.860.321-0.970.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability0.510-1.000.498-1.021.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.274-1.010.126-1.450.6080.332
162Interestvoiceclap_attribute:Interest1.997+0.912.028+0.981.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness-0.063-1.220.042-0.990.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.207-1.290.162-1.470.5230.245
165Longingvoiceclap_attribute:Longing0.510-0.120.444-0.430.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.438-1.140.192-1.850.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive1.514-0.441.253-0.881.7750.594
168Painvoiceclap_attribute:Pain0.254-0.580.190-0.880.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.048-1.51-0.030-1.770.5030.301
170Pridevoiceclap_attribute:Pride0.470-0.890.245-1.530.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.250-0.192.386+0.222.3130.328
172Reliefvoiceclap_attribute:Relief0.636-0.940.488-1.291.0250.415
173Sadnessvoiceclap_attribute:Sadness0.685-0.210.486-0.850.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.627-0.310.529-0.480.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.226-0.98-0.045-1.620.6350.419
176Shamevoiceclap_attribute:Shame0.583+0.200.285-0.850.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.036-0.34-0.024-0.510.1600.360
178Sournessvoiceclap_attribute:Sourness0.367-1.170.324-1.310.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.462-0.450.411-0.590.6230.359
180Teasingvoiceclap_attribute:Teasing0.155-1.020.149-1.030.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude0.883-0.950.548-1.581.3840.529
182Triumphvoiceclap_attribute:Triumph0.302-1.150.330-1.070.7160.359
183Valencevoiceclap_attribute:Valence-0.184-0.28-0.022-0.02-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached2.369+1.312.469+1.491.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold-0.116-0.13-0.031+0.09-0.0660.400
186durationvoiceclap_attribute:duration11.267+0.0612.371+0.2310.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.608+0.733.517+0.553.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.680+0.174.747+0.624.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.969+0.812.948+0.762.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.938+0.681.928+0.531.8910.069
191talking speedvoiceclap_attribute:talking_speed17.690+2.3116.371+1.969.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.10450.09700.0977
1-0.0435-0.0225-0.0852
2-0.0829-0.0832-0.1006
30.08070.11090.0690
40.12440.09430.0501
5-0.0676-0.0773-0.1185
6-0.1059-0.1153-0.0930
70.19200.19850.1698
80.15680.10060.1071
90.08340.08320.1127
100.07890.07690.1265
110.06530.08050.1608
12-0.0372-0.0512-0.0331
13-0.0928-0.0849-0.1013
14-0.0229-0.0254-0.0290
150.18200.19940.1032
16-0.1613-0.1395-0.1271
170.14780.10240.1161
18-0.0182-0.0050-0.0201
190.0329-0.0158-0.0327
200.09460.0894-0.0008
210.03580.01860.0123
22-0.1050-0.0903-0.0965
23-0.04750.01390.0411
24-0.0061-0.0449-0.0183
25-0.0295-0.0288-0.0406
26-0.0982-0.1252-0.0963
27-0.0137-0.0653-0.0552
28-0.0179-0.0353-0.0520
290.05750.07420.0665
300.07750.07820.1683
31-0.0545-0.0427-0.0947
320.07570.06530.1501
33-0.0401-0.0098-0.0271
34-0.0075-0.0353-0.0329
350.11630.12070.0942
36-0.1361-0.1225-0.1103
37-0.0428-0.0434-0.0889
38-0.0847-0.0452-0.0847
39-0.0388-0.0693-0.0594
400.14780.12430.1368
41-0.0367-0.0662-0.0217
420.0058-0.0478-0.0066
430.10090.08990.1175
440.20550.19310.1456
45-0.1163-0.1127-0.0357
460.08290.10370.0967
47-0.0996-0.1144-0.1038
48-0.0123-0.0168-0.0353
49-0.0502-0.0532-0.0509
50-0.0595-0.0849-0.0714
51-0.0838-0.0639-0.0755
520.04280.08900.0232
53-0.0518-0.0697-0.0402
54-0.0960-0.0617-0.0787
55-0.0748-0.0675-0.0987
560.03510.03290.0170
57-0.0451-0.0791-0.1125
580.02710.04290.0978
59-0.0676-0.0532-0.1030
60-0.0969-0.1296-0.0989
610.19020.21370.2139
62-0.0489-0.0402-0.0358
630.06670.06170.0360
64-0.1099-0.1180-0.0859
65-0.0568-0.0608-0.0837
66-0.1397-0.1350-0.1420
67-0.1023-0.0894-0.0946
680.11990.13410.1384
69-0.1072-0.0939-0.0940
700.19110.13680.1465
71-0.0534-0.0469-0.0798
72-0.0771-0.0581-0.0993
73-0.0453-0.1144-0.0445
74-0.0453-0.0249-0.1121
750.05140.05590.0593
76-0.0462-0.0382-0.0222
77-0.01780.0030-0.0938
780.04620.04870.0571
79-0.0784-0.0791-0.0890
800.00080.02740.0168
81-0.0978-0.1015-0.0930
82-0.0964-0.1051-0.0982
830.08340.04290.0140
84-0.0743-0.1006-0.0770
85-0.0516-0.0572-0.0680
860.09240.1359-0.0167
87-0.0514-0.0552-0.0195
880.14600.14840.1113
89-0.00370.0101-0.0194
90-0.1460-0.1279-0.1113
910.00430.0249-0.0008
92-0.0694-0.0492-0.0930
930.06800.04940.0829
94-0.0053-0.03820.0685
95-0.1199-0.1153-0.1102
96-0.1325-0.1493-0.1349
97-0.0237-0.0326-0.0224
980.08520.05900.0160
99-0.0229-0.04400.0564
100-0.0543-0.1073-0.0539
101-0.01280.00420.0131
1020.03650.00750.0126
103-0.0888-0.0952-0.1083
104-0.0561-0.0796-0.0701
1050.0027-0.0393-0.0339
106-0.0170-0.0324-0.0087
107-0.0213-0.0121-0.0119
1080.11170.07730.0565
1090.03040.00880.0157
110-0.01080.03310.0452
111-0.1140-0.1068-0.1190
112-0.1104-0.1037-0.1011
113-0.0825-0.0800-0.0327
114-0.0865-0.0729-0.0898
1150.19200.22890.2644
116-0.0852-0.0885-0.0880
1170.0064-0.00240.0174
1180.02660.01510.0033
119-0.0991-0.0966-0.0992
120-0.0906-0.1001-0.1055
121-0.0109-0.03580.0128
122-0.0415-0.0380-0.0730
1230.09280.14130.1311
1240.0030-0.0250-0.0240
1250.22980.16990.0850
1260.02160.05250.0869
1270.08470.04650.0798
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.0831-0.0677-0.0550
10.01790.04150.0176
20.05270.07840.0210
30.01840.04890.0292
40.01140.0084-0.0455
50.02260.0156-0.0064
60.01230.06490.0355
7-0.0268-0.0591-0.0506
8-0.03500.00850.0321
9-0.0035-0.0191-0.0640
10-0.0068-0.0150-0.0423
110.03240.04230.0589
12-0.0556-0.0484-0.0547
13-0.0382-0.0593-0.0441
14-0.0942-0.0769-0.0714
150.09280.10490.0621
160.06330.04990.1235
170.02720.02430.0023
180.0138-0.0092-0.0292
19-0.1305-0.0927-0.0410
200.03380.03540.0098
21-0.0527-0.0599-0.0464
220.08890.08710.1545
23-0.0050-0.01380.0086
240.11450.07230.0436
25-0.0913-0.0718-0.0919
260.03360.01680.0994
270.14110.14360.1273
280.07730.10390.0614
290.03030.0325-0.0062
300.06060.0446-0.0131
310.0256-0.0083-0.0484
32-0.0043-0.0481-0.0274
330.05610.01630.0195
340.0178-0.00870.0168
35-0.1189-0.0876-0.0644
360.00750.0379-0.0210
370.01660.0003-0.0066
38-0.0633-0.0639-0.0651
39-0.0185-0.0413-0.0138
40-0.0242-0.0146-0.0105
410.0064-0.00690.0626
42-0.0573-0.0637-0.0497
43-0.0396-0.0621-0.0163
440.03070.04000.0220
45-0.02140.0083-0.0502
46-0.0255-0.0312-0.0734
470.11020.11870.1138
480.06110.0657-0.0003
490.11070.08250.0443
50-0.0056-0.0171-0.0160
510.0119-0.00600.0298
52-0.0266-0.0266-0.0493
530.19230.15180.1462
540.01400.02610.0297
550.21360.26690.2124
560.00300.00020.0464
570.03010.04560.0594
58-0.0399-0.0606-0.0443
59-0.0387-0.0784-0.0553
600.02050.02670.0768
61-0.0532-0.0621-0.0464
620.01330.01820.0563
630.06330.09070.0457
640.03460.01730.0278
650.05360.07640.0320
660.0379-0.00430.0158
670.07680.10800.1609
680.03530.01380.0483
690.03030.04230.0096
70-0.0500-0.0621-0.0724
710.09960.11770.0933
72-0.0691-0.0683-0.0554
73-0.01560.00010.0069
74-0.0459-0.0331-0.0269
750.11070.08910.1067
760.0184-0.0038-0.0476
770.0045-0.02550.0328
780.0291-0.0199-0.0078
790.0242-0.00150.0119
80-0.0577-0.0573-0.0531
810.03120.00160.0363
820.07680.06340.1132
83-0.0290-0.0560-0.0444
840.01850.02220.0307
850.04640.01180.0299
86-0.0214-0.0138-0.0476
87-0.0372-0.0104-0.0426
880.07640.07080.0359
890.12130.10800.1140
900.02810.04790.0400
910.0205-0.0241-0.0628
920.0014-0.0120-0.0287
930.13140.12530.1112
940.0064-0.0287-0.0140
950.00130.00210.0602
960.04370.00550.0206
97-0.0288-0.0367-0.0226
980.0060-0.0227-0.0284
990.00530.00700.0468
100-0.01120.0068-0.0602
1010.14790.17830.1555
1020.09280.06930.0630
103-0.0483-0.0418-0.0327
104-0.0602-0.04180.0019
105-0.0706-0.0677-0.0694
106-0.0541-0.0428-0.0411
1070.02200.01360.0311
1080.08460.12430.1508
1090.02280.01400.0923
110-0.0483-0.01480.0092
1110.10340.13650.1504
112-0.0358-0.0309-0.0400
1130.02050.0127-0.0313
1140.05870.01520.0635
1150.0076-0.0351-0.0378
116-0.0266-0.0111-0.0066
1170.07440.07440.0615
118-0.0152-0.0571-0.0696
1190.03700.04710.0547
1200.07880.07030.1071
121-0.0515-0.0599-0.0550
1220.01530.01320.0070
1230.17590.13960.1593
1240.0012-0.04080.0366
1250.05940.05880.0151
1260.05820.0578-0.0211
127-0.0205-0.0423-0.0549
1280.05750.03870.0646
129-0.0108-0.0183-0.0157
130-0.0783-0.0790-0.0536
1310.06520.03090.0411
1320.12660.12480.0676
1330.09330.06520.0459
1340.06620.10650.0545
1350.0370-0.0032-0.0127
1360.05800.08200.1052
1370.13920.08200.1157
1380.01880.04560.0349
139-0.0587-0.0596-0.0416
140-0.0382-0.0134-0.0180
141-0.0967-0.0591-0.0528
142-0.0377-0.0461-0.0461
143-0.00050.02310.0301
144-0.0208-0.0479-0.0556
145-0.0435-0.0540-0.0294
1460.02680.04410.0548
1470.0208-0.02410.0241
1480.06770.05200.0992
1490.0126-0.00130.0410
150-0.00510.0031-0.0410
151-0.0326-0.0073-0.0392
1520.07440.10390.0329
1530.04040.02310.0482
1540.09960.04430.0809
1550.02320.00060.0077
1560.0157-0.0367-0.0085
1570.02020.01500.0236
1580.00400.01340.0252
159-0.0735-0.0973-0.0625
160-0.0081-0.0438-0.0507
161-0.0628-0.0548-0.0120
162-0.0309-0.0341-0.0506
163-0.1213-0.0953-0.0594
164-0.0797-0.0739-0.0469
1650.07060.03820.0014
166-0.0154-0.0087-0.0089
1670.04420.07540.0770
168-0.0822-0.0881-0.0754
169-0.0214-0.05350.0367
1700.04760.0237-0.0113
171-0.00770.03130.0129
172-0.0389-0.0525-0.0618
1730.01870.04410.0791
174-0.0865-0.0718-0.0548
175-0.0696-0.0713-0.0859
1760.07390.08560.1174
177-0.0343-0.0234-0.0440
1780.0022-0.0039-0.0474
1790.07490.04790.1012
180-0.0281-0.0402-0.0611
181-0.0274-0.0349-0.0715
182-0.0355-0.0558-0.0461
1830.02330.03410.0547
184-0.0524-0.0530-0.0472
1850.04910.06010.0179
186-0.0404-0.0349-0.0477
1870.04300.08300.0237
1880.06620.14160.1471
189-0.0372-0.0474-0.0016
1900.04860.05250.0386
1910.15560.15080.1680
192-0.0602-0.0677-0.0646
1930.0049-0.0033-0.0562
194-0.0423-0.0325-0.0692
195-0.0115-0.0413-0.0507
1960.00250.04580.0467
1970.05360.0039-0.0746
1980.02950.02160.0235
1990.08410.06880.0786
200-0.0851-0.1009-0.0617
2010.04540.09420.0622
202-0.0739-0.0800-0.0797
2030.08220.04100.1179
2040.20390.19150.1963
2050.10540.10390.0665
2060.04660.0400-0.0550
207-0.0312-0.02920.0070
2080.02530.00100.0302
2090.08990.09070.0443
2100.07440.05220.0226
211-0.0309-0.0334-0.0251
2120.03380.02600.0334
2130.09960.06320.0512
2140.07440.04530.0223
215-0.0367-0.0349-0.0448
216-0.1005-0.0683-0.0638
217-0.0510-0.0070-0.0379
2180.06190.0815-0.0238
219-0.0764-0.0937-0.0545
2200.03650.03540.1053
221-0.00450.00160.0242
222-0.0298-0.0349-0.0346
2230.09570.09420.0358
2240.0071-0.00330.0062
2250.13140.09930.1349
2260.07970.08560.0475
227-0.0358-0.0499-0.0698
228-0.0291-0.0103-0.0326
2290.05580.07690.0230
230-0.0529-0.03950.0145
231-0.01570.03340.0092
232-0.0619-0.0132-0.0378
2330.06810.04080.0155
2340.17980.16500.1700
2350.01800.05020.0181
236-0.0457-0.0583-0.0458
2370.02670.0446-0.0279
2380.02280.0400-0.0320
239-0.0592-0.0476-0.0551
240-0.0677-0.0759-0.0544
2410.05770.04920.0415
242-0.00370.03640.0350
243-0.00090.0034-0.0195
244-0.0411-0.0820-0.0567
2450.08020.0749-0.0419
246-0.00070.00810.0188
247-0.0476-0.0362-0.0449
248-0.0193-0.0304-0.0610
249-0.0623-0.0266-0.0528
Source: Emolia / Emilia source archive · SHA-256 ad5351ce396e0e93…
e02GermanEmolia / Emilia

Emolia DE · DE_B00000_S00170_W000025

Natural speech · Original speech · 12.4 s

Wie sehen die Zahlen, Entschuldigung, ich meine Menschen mit Hoffnungen, Träumen und Familien, am Ende dieser Tabellen aus? Bell hatte die Firma im letzten Oktober verlassen, nachdem er sich nach eigenen Angaben versichert hatte, dass die Zukunft des Studios jetzt gesichert sei.

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Disappointment
z +1.38 / +2.09raw 0.97 / 1.46
Pride
z +1.50 / +1.12raw 0.76 / 0.57
Sadness
z +0.80 / +1.44raw 0.61 / 1.09
Contentment
z +1.39 / +1.08raw 0.71 / 0.55
Concentration
z +1.17 / +0.98raw 1.50 / 1.36
Relief
z +0.51 / +1.10raw 0.38 / 0.78
Bitterness
z +0.80 / +1.09raw 0.35 / 0.47

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Newsreader style (S_NEWS)
z +4.81 / +4.87raw 3.14 / 3.17
Formal style (S_FORM)
z +2.33 / +2.29raw 3.57 / 3.54
Teacher or didactic style (S_TECH)
z +2.26 / +2.32raw 3.42 / 3.50
Vocal register (REGS)
z +2.04 / +2.10raw 3.79 / 3.87
Authoritative style (S_AUTH)
z +2.02 / +2.06raw 3.19 / 3.24
Speech structure (STRU)
z +1.93 / +1.93raw 3.99 / 3.99
Articulation clarity (CLRT)
z +1.87 / +1.90raw 3.79 / 3.82

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z -1.10 / -1.12raw 0.55 / 0.52
Vocal-burst blend (0–10)z -2.39 / -2.30raw 0.13 / 0.38
Content enjoyment (CE)z +0.82 / +0.94raw 5.63 / 5.76
Production quality (PQ)z +0.34 / +0.51raw 7.35 / 7.49
Speech signal quality, calibratedz +0.25 / +0.19raw 3.61 / 3.58
Background quality, calibratedz +0.63 / +0.47raw 3.98 / 3.83
Overall quality, calibratedz +0.74 / +0.62raw 3.29 / 3.22
P.808 MOS estimatez +0.79 / +1.04raw 3.81 / 3.94
score content enjoymentz +1.16 / +1.22raw 4.86 / 4.88
score overall qualityz +0.60 / +0.75raw 2.96 / 3.03
Background Noisez -1.67 / -2.00raw 0.37 / 0.28
Tempo (TEMP)z +1.18 / +1.38raw 3.52 / 3.76
Arousal (AROU)z +0.49 / +0.48raw 2.92 / 2.91
Valence (VALN)z +0.45 / +0.56raw 2.83 / 2.96
Recording quality (RCQL)z +1.53 / +1.33raw 4.11 / 3.93
Background noise (BKGN)z +1.88 / +1.69raw 4.14 / 3.99
talking speedz +1.43 / +1.67raw 14.41 / 15.29
Characters per second (CPS): 22.45 raw · z +3.57

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8510.877
Identity · 250D0.7350.820

Vocal-burst regions

Tiny
Base
0 s12.4 s

Verified burst intervals are unavailable for this clip.

Tiny: 0 regions, count-head estimate 0.01; Base: 0 regions, count-head estimate 0.01.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.434+0.750.250+0.430.0070.570
1Amusementemo_Amusement0.178+0.200.115+0.13-0.0010.885
2Angeremo_Anger0.320+0.520.338+0.550.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.080+0.130.170+0.290.0050.572
4Aweemo_Awe0.006+0.010.014+0.030.0000.427
5Bitternessemo_Bitterness0.346+0.800.467+1.090.0130.415
6Concentrationemo_Concentration1.504+1.171.364+0.980.6340.741
7Confusionemo_Confusion0.094+0.170.413+0.760.0010.540
8Contemplationemo_Contemplation0.677+0.990.731+1.090.0820.598
9Contemptemo_Contempt0.267+0.580.216+0.47-0.0000.464
10Contentmentemo_Contentment0.714+1.390.554+1.080.0000.514
11Disappointmentemo_Disappointment0.966+1.381.460+2.090.0000.700
12Disgustemo_Disgust0.240+0.480.114+0.220.0070.489
13Distressemo_Distress0.407+0.500.694+0.840.0000.821
14Doubtemo_Doubt0.066+0.130.259+0.540.0060.466
15Elationemo_Elation0.463+0.620.197+0.250.0120.728
16Embarrassmentemo_Embarrassment0.103+0.140.237+0.450.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.176+0.190.193+0.210.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.027-0.000.042+0.030.0280.539
19Fearemo_Fear-0.001-0.030.122+0.180.0150.610
20Helplessnessemo_Helplessness0.278+0.410.436+0.63-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.924+1.060.624+0.670.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.440+0.410.501+0.500.1490.705
23Infatuationemo_Infatuation-0.034-0.04-0.005+0.02-0.0150.533
24Interestemo_Interest2.392+0.892.217+0.691.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness-0.052-0.05-0.019+0.01-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.167+0.240.304+0.550.0590.442
27Longingemo_Longing0.182+0.150.271+0.290.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.218+0.300.138+0.150.0550.534
29Painemo_Pain0.005+0.010.036+0.06-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.231+0.350.092+0.120.0170.614
31Prideemo_Pride0.765+1.500.570+1.120.0020.509
32Reliefemo_Relief0.378+0.510.779+1.100.0330.681
33Sadnessemo_Sadness0.608+0.801.089+1.440.0000.759
34Sexual Lustemo_Sexual_Lust-0.004-0.07-0.031-0.120.0300.494
35Shameemo_Shame0.208+0.380.514+0.980.0130.510
36Sournessemo_Sourness0.258+0.600.387+0.89-0.0050.439
37Teasingemo_Teasing0.142+0.260.335+0.610.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.401+0.520.350+0.440.0410.697
39Triumphemo_Triumph0.284+0.560.189+0.38-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg3.223+0.192.944-0.032.9841.238
41Arousal (AROU)vn_AROU_reg2.919+0.492.910+0.482.3591.147
42Arousal shift (ARSH)vn_ARSH_reg3.621+1.203.587+1.162.5800.867
43Vocal attack (ATCK)vn_ATCK_reg3.060+0.693.405+0.962.1621.295
44Background noise (BKGN)vn_BKGN_reg4.136+1.883.988+1.692.7030.761
45Brightness (BRGT)vn_BRGT_reg3.306+1.623.300+1.611.9910.813
46Speech chunking (CHNK)vn_CHNK_reg3.674+1.483.830+1.601.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg3.786+1.873.816+1.902.0180.945
48Cognitive load (COGL)vn_COGL_reg0.918-1.110.770-1.252.0941.061
49Dynamic arc (DARC)vn_DARC_reg2.437+0.602.432+0.591.6151.380
50Disfluency (DFLU)vn_DFLU_reg0.979-1.270.665-1.542.4881.184
51Emphasis (EMPH)vn_EMPH_reg3.123+0.343.068+0.302.6071.524
52Esthetics (ESTH)vn_ESTH_reg3.015+0.783.058+0.842.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.067-0.060.022-0.200.0860.322
54Focus (FOCS)vn_FOCS_reg3.446+1.003.342+0.912.3441.102
55Fullness (FULL)vn_FULL_reg3.125+1.513.207+1.611.7780.889
56Perceived gender (GEND)vn_GEND_reg1.256-1.761.368-1.694.0551.588
57Harmonicity (HARM)vn_HARM_reg3.552+1.283.280+0.962.4800.836
58Metallic character (METL)vn_METL_reg2.118+0.522.275+0.681.6210.958
59Pitch range (RANG)vn_RANG_reg3.393+0.533.487+0.642.9140.897
60Recording quality (RCQL)vn_RCQL_reg4.109+1.533.933+1.332.7540.888
61Vocal register (REGS)vn_REGS_reg3.788+2.043.866+2.101.1981.268
62Respiration (RESP)vn_RESP_reg1.848-0.791.815-0.822.6801.056
63Roughness (ROUG)vn_ROUG_reg1.968-0.071.903-0.152.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg2.484+0.412.701+0.672.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg2.819+0.822.825+0.822.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg2.858+0.952.794+0.881.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg2.874+1.712.658+1.451.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg2.817+0.022.734-0.052.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg2.539+0.482.513+0.442.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg2.842+0.962.798+0.912.0700.803
71Smoothness (SMTH)vn_SMTH_reg3.382+1.133.547+1.292.1431.093
72Stance (STNC)vn_STNC_reg3.352+0.633.486+0.752.6331.137
73Speech structure (STRU)vn_STRU_reg3.986+1.933.986+1.932.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg-0.083-1.03-0.363-1.231.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg3.192+2.023.241+2.061.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg1.268-0.211.270-0.211.5601.401
77Casual style (S_CASU)vn_S_CASU_reg3.195-0.473.275-0.413.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg3.161+0.433.183+0.452.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg3.280+0.493.126+0.352.7211.148
80Formal style (S_FORM)vn_S_FORM_reg3.569+2.333.538+2.291.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg3.223+0.313.190+0.292.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg2.933+1.122.647+0.931.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg3.139+4.813.174+4.870.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg2.858+0.383.132+0.592.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg1.865+0.512.152+0.731.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg2.658+0.132.747+0.202.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg3.421+2.263.497+2.320.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg-0.469-1.28-0.868-1.521.5971.618
89Tempo (TEMP)vn_TEMP_reg3.522+1.183.756+1.382.1171.190
90Tension (TENS)vn_TENS_reg2.098+0.631.929+0.441.5590.852
91Valence (VALN)vn_VALN_reg2.835+0.452.958+0.562.3281.118
92Valence shift (VALS)vn_VALS_reg3.180+0.613.189+0.622.5181.079
93Velocity flux (VFLX)vn_VFLX_reg3.070+0.673.287+0.852.2561.208
94Volatility (VOLT)vn_VOLT_reg1.762-0.931.877-0.802.5860.886
95Vulnerability (VULN)vn_VULN_reg1.726-0.941.664-1.002.6480.984
96Warmth (WARM)vn_WARM_reg2.712+0.152.690+0.122.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_60.553-1.100.516-1.122.3321.621
98Vocal-burst blend (0–10)blend_0_100.125-2.390.380-2.306.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.117-2.900.133-2.851.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.312+0.213.229-0.073.2500.293
101Arousaleiv_extra:Arousal1.548+0.601.510+0.511.2890.435
102Authenticityeiv_extra:Authenticity3.002+2.192.969+1.982.6670.153
103Background Noiseeiv_extra:Background_Noise0.367-1.670.278-2.000.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.014-0.081.210+0.561.0400.303
105Gendereiv_extra:Gender1.130+1.751.294+1.92-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.789-0.151.707-0.481.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive2.271+0.882.311+0.961.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.757+1.852.731+1.772.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous0.978+0.140.944+0.040.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.481+0.620.509+0.710.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.632+0.790.392+0.050.3760.324
112Valenceeiv_extra:Valence0.159+0.41-0.154-0.420.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.785+0.161.505-0.481.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold0.186+0.450.110+0.200.0510.300
115score background qualityeiv_extra:score_background_quality3.553+0.363.621+0.463.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.859+1.164.878+1.224.5070.303
117score overall qualityeiv_extra:score_overall_quality2.957+0.603.030+0.752.6520.507
118score speech qualityeiv_extra:score_speech_quality1.879+0.491.887+0.541.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.629+0.825.757+0.944.7991.018
120Content usefulness (CU)audiobox:CU6.120+0.476.113+0.465.6491.001
121Production complexity (PC)audiobox:PC1.707-0.391.719-0.382.3311.617
122Production quality (PQ)audiobox:PQ7.346+0.347.493+0.517.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw4.134+0.354.087+0.283.9000.669
124Background quality, rawdnsmos:BAK_raw4.135+0.823.933+0.643.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.773+0.853.691+0.753.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.612+0.253.582+0.193.4880.501
127Background quality, calibrateddnsmos:BAK3.977+0.633.829+0.473.3890.928
128Overall quality, calibrateddnsmos:OVRL3.288+0.743.217+0.622.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.814+0.793.938+1.043.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.013+0.020.010+0.020.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection-0.002-1.710.673-0.490.9470.554
132Agevoiceclap_attribute:Age3.476+0.453.244+0.033.2290.547
133Amusementvoiceclap_attribute:Amusement0.292-0.180.478+0.150.3920.572
134Angervoiceclap_attribute:Anger0.545-0.880.629-0.720.9920.506
135Arousalvoiceclap_attribute:Arousal1.087-0.531.610+0.311.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.383-0.260.874+0.760.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.948+1.072.914+0.862.7700.167
138Awevoiceclap_attribute:Awe0.160-1.460.266-1.050.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.278-1.100.228-1.290.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.381-0.900.483-0.460.5920.236
141Concentrationvoiceclap_attribute:Concentration1.778+1.061.222+0.021.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.083+0.031.190+0.261.0690.468
143Confusionvoiceclap_attribute:Confusion0.380-0.790.609-0.100.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.752-0.070.499-0.690.7800.406
145Contemptvoiceclap_attribute:Contempt0.655-0.160.702-0.020.7090.339
146Contentmentvoiceclap_attribute:Contentment0.345-0.790.486-0.330.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.510-0.860.615-0.500.7600.291
148Disgustvoiceclap_attribute:Disgust0.468-0.760.651+0.130.6250.207
149Distressvoiceclap_attribute:Distress0.122-2.300.459-1.340.9270.350
150Doubtvoiceclap_attribute:Doubt0.344-1.020.340-1.040.5350.188
151Elationvoiceclap_attribute:Elation0.352-0.470.760+0.560.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.094-0.680.413+0.340.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.470-0.730.561-0.390.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.010-1.730.177-1.090.4620.261
155Fearvoiceclap_attribute:Fear0.335-1.330.343-1.300.6180.212
156Gendervoiceclap_attribute:Gender1.136+1.281.262+1.41-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.141-1.930.173-1.810.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.704-0.361.659-0.481.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.604-0.260.996+0.730.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability0.413-1.180.737-0.581.0520.541
161Infatuationvoiceclap_attribute:Infatuation-0.188-2.400.291-0.950.6080.332
162Interestvoiceclap_attribute:Interest2.268+1.522.199+1.371.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness-0.012-1.110.213-0.620.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.513-0.040.525+0.010.5230.245
165Longingvoiceclap_attribute:Longing0.064-2.160.246-1.330.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.373-1.330.484-1.010.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive1.954+0.302.282+0.851.7750.594
168Painvoiceclap_attribute:Pain-0.045-1.960.159-1.020.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.079-1.410.431-0.240.5030.301
170Pridevoiceclap_attribute:Pride0.551-0.660.586-0.560.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.881+1.732.782+1.432.3130.328
172Reliefvoiceclap_attribute:Relief0.446-1.400.759-0.641.0250.415
173Sadnessvoiceclap_attribute:Sadness0.261-1.570.386-1.170.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.724-0.141.087+0.500.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust-0.042-1.620.111-1.250.6350.419
176Shamevoiceclap_attribute:Shame0.099-1.510.335-0.680.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.185+0.070.292+0.370.1600.360
178Sournessvoiceclap_attribute:Sourness0.631-0.260.740+0.120.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.452-0.480.521-0.290.6230.359
180Teasingvoiceclap_attribute:Teasing0.544+0.010.753+0.560.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude0.519-1.630.722-1.251.3840.529
182Triumphvoiceclap_attribute:Triumph0.238-1.330.514-0.560.7160.359
183Valencevoiceclap_attribute:Valence0.343+0.550.143+0.24-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached2.295+1.181.758+0.211.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold0.161+0.570.307+0.93-0.0660.400
186durationvoiceclap_attribute:duration11.841+0.1513.986+0.4810.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.671+0.853.776+1.073.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment5.087+2.935.126+3.204.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality3.145+1.243.196+1.362.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.945+0.781.973+1.181.8910.069
191talking speedvoiceclap_attribute:talking_speed14.406+1.4315.286+1.679.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.18720.16800.0958
10.13700.15450.1486
20.01230.04200.1549
30.09450.09220.1254
4-0.0730-0.0525-0.0745
50.0103-0.0589-0.0386
60.01070.04670.0729
70.02020.08890.0725
80.0472-0.01470.0322
9-0.0541-0.0016-0.0375
10-0.0758-0.0836-0.0736
11-0.0772-0.0600-0.0554
12-0.0730-0.0671-0.0797
130.07530.05250.1009
14-0.1137-0.0922-0.0825
150.19900.21150.2020
16-0.0886-0.06190.0582
17-0.0625-0.0444-0.0823
18-0.0854-0.0844-0.0731
19-0.0283-0.0242-0.0622
20-0.0100-0.0433-0.0106
210.14330.21000.1520
220.10320.09520.0731
23-0.0557-0.07390.0747
240.13700.11250.0708
250.03650.04330.0409
26-0.0730-0.0907-0.0743
270.03720.02210.0193
28-0.0986-0.0975-0.0873
29-0.1187-0.1080-0.0848
300.04630.05700.0722
31-0.0735-0.0675-0.0731
320.07400.11550.0919
330.03010.04610.1200
34-0.0835-0.0832-0.0710
350.03010.0341-0.0366
36-0.0418-0.0536-0.0512
370.0162-0.00010.0377
380.08490.08660.0740
390.07350.08620.0726
40-0.0088-0.0433-0.0764
410.06160.08920.0466
42-0.1128-0.0754-0.0799
43-0.0285-0.0596-0.0300
44-0.03330.0215-0.0475
45-0.1155-0.0780-0.1065
46-0.0170-0.0031-0.0073
470.00340.03770.0214
48-0.0931-0.0907-0.0773
49-0.01500.0092-0.0090
50-0.0881-0.0806-0.0298
510.10680.07160.0351
52-0.0607-0.0720-0.0554
53-0.0945-0.0960-0.0808
54-0.1625-0.1072-0.0886
55-0.0251-0.0067-0.0580
56-0.1087-0.0761-0.0751
570.07580.06110.0331
58-0.02090.00880.0065
590.05270.06410.1196
600.01390.0010-0.0764
610.13150.15900.1292
620.07530.05290.0689
63-0.0518-0.01180.0373
640.11780.08510.1320
65-0.0571-0.0559-0.0646
66-0.01890.04050.0060
670.01500.02300.0501
680.08720.10350.1830
690.04630.07350.1824
700.22370.17850.0773
71-0.0703-0.0626-0.0878
720.23370.26090.2448
730.06570.04780.0569
74-0.0156-0.0119-0.0615
750.12970.15000.1654
76-0.03040.0017-0.0481
77-0.1023-0.0904-0.0723
78-0.0612-0.0889-0.0705
790.14880.06190.1211
80-0.0931-0.1237-0.0717
810.04360.08960.1636
82-0.0441-0.0525-0.0124
83-0.1077-0.0784-0.0811
840.03740.05210.0010
85-0.0694-0.0915-0.0841
860.09080.19120.1862
870.00470.00020.0426
88-0.0963-0.0952-0.0735
89-0.0849-0.0885-0.0727
900.13420.10800.0870
91-0.0653-0.0626-0.0728
920.00000.01060.0154
93-0.0589-0.0276-0.0774
940.17800.13050.1378
95-0.0475-0.0381-0.0386
96-0.1461-0.1717-0.0617
97-0.0128-0.02610.0136
98-0.0991-0.0791-0.0760
990.09680.12670.0636
1000.0128-0.0186-0.0070
101-0.0313-0.0506-0.0494
1020.06940.04720.0984
1030.05090.02760.1089
1040.11780.10420.1504
105-0.0639-0.0369-0.0764
106-0.0225-0.0300-0.0280
107-0.1105-0.0671-0.0738
108-0.1087-0.1177-0.0859
1090.29030.31340.2322
110-0.0735-0.0347-0.1019
111-0.0308-0.0352-0.0282
112-0.0260-0.01490.0134
113-0.02990.0042-0.0287
1140.01960.0255-0.0268
1150.03930.0309-0.0052
1160.06440.07120.1037
1170.0227-0.01190.0194
118-0.0931-0.0757-0.0771
119-0.0831-0.0517-0.0766
1200.08170.16350.1370
121-0.0525-0.0206-0.0260
122-0.0972-0.1072-0.0814
1230.02450.0160-0.0404
1240.08990.07010.1125
1250.09080.0266-0.0361
1260.09540.07720.1259
127-0.0374-0.0469-0.0840
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.0526-0.0827-0.0514
1-0.00820.01660.0226
20.00010.00970.0453
3-0.1299-0.1482-0.0887
4-0.0031-0.0105-0.0278
50.07010.06710.0020
6-0.00620.0205-0.0397
70.0070-0.0354-0.0473
8-0.0721-0.0634-0.0731
90.02490.04050.0307
100.00810.0282-0.0490
11-0.0270-0.0457-0.0400
120.05670.06850.0653
130.13870.21700.1194
14-0.0519-0.0424-0.0130
150.06730.09290.0658
16-0.01060.0006-0.0756
17-0.0748-0.0832-0.0561
18-0.00010.02740.0780
19-0.1121-0.0714-0.0742
20-0.0646-0.0354-0.0468
21-0.0388-0.0337-0.0433
220.00170.01250.0087
230.01410.00680.0072
240.14900.10150.0295
25-0.0194-0.00490.0531
260.10730.14180.1169
270.05910.03180.0515
280.01680.0006-0.0368
290.0308-0.0035-0.0450
30-0.0384-0.0478-0.0406
31-0.0331-0.0491-0.0587
320.09840.10580.0351
330.05600.04110.0629
340.0285-0.0101-0.0558
35-0.0830-0.0623-0.0602
36-0.0098-0.0052-0.0140
370.09160.05290.0301
38-0.0998-0.1117-0.0648
390.02140.05240.0956
40-0.01830.04780.0728
41-0.0523-0.0489-0.0536
420.10250.11440.0581
43-0.04510.0095-0.0552
44-0.0256-0.0376-0.0562
45-0.0287-0.0333-0.0499
460.01890.0244-0.0133
47-0.0219-0.0159-0.0271
48-0.01910.0070-0.0148
490.08890.03340.0453
500.10250.13690.1289
510.05430.0057-0.0102
52-0.0448-0.0526-0.0461
530.08100.0275-0.0175
54-0.0059-0.0127-0.0668
550.00150.0077-0.0315
56-0.0308-0.0153-0.0598
570.03640.02220.0126
58-0.0350-0.0448-0.0449
590.16950.15360.0827
600.03200.08490.0882
61-0.03210.01840.0317
620.06050.05450.0833
63-0.0936-0.0529-0.0508
640.02870.00050.0060
65-0.1080-0.1020-0.0570
66-0.0605-0.0653-0.0589
67-0.0820-0.0956-0.0492
680.02490.05910.0574
690.07040.05530.0237
70-0.0871-0.0811-0.0677
71-0.0388-0.0438-0.0490
720.02240.00360.0004
730.0203-0.0203-0.0491
74-0.0649-0.0687-0.0504
75-0.0321-0.0671-0.0680
760.06600.08110.1647
770.0465-0.0238-0.0450
780.04610.05370.0842
79-0.01610.0019-0.0300
80-0.0591-0.0299-0.0496
810.10050.06530.0641
820.04310.02010.0519
830.13330.10100.0515
84-0.0502-0.0725-0.0703
85-0.0236-0.0360-0.0499
86-0.0197-0.0108-0.0365
87-0.0762-0.0924-0.0398
88-0.0666-0.0295-0.0493
890.01770.06530.0272
900.15100.14500.2108
910.01470.01300.0512
92-0.0666-0.0886-0.0431
930.0101-0.0228-0.0566
940.05260.06500.0005
95-0.01180.02470.1181
960.05640.06770.0303
97-0.1422-0.1327-0.0588
980.04610.0142-0.0527
99-0.0496-0.0142-0.0624
1000.12640.14720.1998
1010.01900.03280.0368
1020.04480.05050.0922
1030.07240.08060.0866
104-0.02180.01360.0660
105-0.0731-0.05100.0067
106-0.0468-0.0285-0.0384
1070.16880.16220.1348
108-0.0226-0.0337-0.0371
1090.09430.04460.0377
110-0.0031-0.0044-0.0295
1110.12990.12570.2178
1120.04410.09770.1364
1130.0181-0.0720-0.0366
1140.04990.03810.0311
115-0.0076-0.0279-0.0353
1160.13400.12570.1287
1170.00960.02980.0634
118-0.0383-0.0489-0.0651
1190.08710.07520.1263
120-0.0248-0.0311-0.0276
121-0.0465-0.0137-0.0292
122-0.0475-0.0177-0.0471
1230.01500.0041-0.0515
1240.09230.07840.0102
1250.06390.0505-0.0046
1260.06460.02120.0240
127-0.0425-0.0647-0.0513
1280.03500.07520.0506
129-0.0006-0.0093-0.0529
1300.15450.11600.0806
131-0.0357-0.0569-0.0437
132-0.00670.00230.0295
1330.00860.0188-0.0493
1340.06900.03320.0304
1350.0390-0.0042-0.0199
1360.14010.15900.1769
1370.06900.06870.0737
138-0.0625-0.0698-0.0440
1390.05980.10150.1186
1400.07930.05610.0576
1410.03140.02930.0273
1420.0020-0.0346-0.0431
143-0.0680-0.0478-0.0231
1440.01660.00080.0657
145-0.0653-0.0779-0.0634
146-0.0302-0.0109-0.0339
147-0.0066-0.0392-0.0443
1480.06250.0612-0.0082
1490.12370.07520.1094
150-0.0711-0.0741-0.0491
151-0.0349-0.01840.0489
152-0.0567-0.0435-0.0553
1530.0313-0.0152-0.0244
1540.0186-0.0457-0.0486
1550.14220.09240.0673
1560.03860.0139-0.0500
1570.01380.0286-0.0194
158-0.0212-0.0655-0.0107
159-0.0557-0.0892-0.0584
1600.10660.04380.0170
1610.06220.02200.0233
1620.0117-0.0132-0.0323
163-0.0455-0.01760.0396
1640.0059-0.0009-0.0438
165-0.0163-0.0251-0.0241
166-0.0519-0.0532-0.0580
167-0.00390.07040.0728
168-0.0687-0.1278-0.0705
1690.10050.09830.0166
1700.00350.0105-0.0099
1710.0042-0.00760.0929
172-0.1476-0.1772-0.0578
173-0.01270.01460.0035
174-0.0359-0.0397-0.0512
175-0.0376-0.0626-0.0689
176-0.0502-0.0119-0.0424
177-0.0649-0.0682-0.0411
178-0.01630.0030-0.0371
1790.00700.0055-0.0614
1800.07930.06610.0266
181-0.0299-0.0103-0.0168
182-0.0455-0.0532-0.0431
183-0.0759-0.0763-0.0822
184-0.00740.01110.0284
1850.06460.06070.0312
1860.0261-0.0011-0.0276
1870.00450.0134-0.0368
1880.0051-0.0231-0.0149
189-0.0636-0.0048-0.0169
1900.03540.01560.0496
1910.0347-0.00480.1057
192-0.0174-0.0074-0.0604
193-0.0289-0.0419-0.0526
194-0.0827-0.0674-0.0647
195-0.0513-0.0459-0.0474
1960.02510.02320.0270
197-0.0100-0.0642-0.0698
198-0.0367-0.0310-0.0482
199-0.0639-0.1096-0.0548
200-0.0413-0.0634-0.0576
201-0.0581-0.0243-0.0470
202-0.0150-0.0360-0.1038
203-0.0342-0.0363-0.0622
2040.05710.0344-0.0274
2050.04920.02600.0638
206-0.0284-0.0526-0.0514
207-0.0752-0.1004-0.0674
2080.02560.00360.0223
2090.0198-0.0146-0.0276
2100.11480.12410.0812
211-0.0224-0.0316-0.0383
2120.07210.05960.0909
2130.10180.06930.0731
2140.06630.04110.0292
215-0.0560-0.0422-0.0419
2160.10660.08060.1686
2170.00610.05400.0914
2180.00040.02670.0261
219-0.0224-0.0709-0.0649
220-0.0998-0.0779-0.0551
2210.05570.11870.0386
222-0.0301-0.0104-0.0585
223-0.0601-0.0588-0.0400
224-0.0048-0.0329-0.0455
2250.18860.11390.0382
226-0.01300.0138-0.0236
227-0.0772-0.0784-0.0659
228-0.0509-0.0757-0.0808
229-0.0601-0.0200-0.0131
2300.0161-0.01640.0224
2310.07280.03760.0455
232-0.0748-0.0768-0.0685
233-0.0704-0.0779-0.0659
234-0.0106-0.0247-0.0423
2350.0001-0.0193-0.0410
236-0.0203-0.0340-0.0386
2370.02600.02670.0487
2380.02050.01960.0408
239-0.0584-0.0502-0.0515
240-0.0584-0.0908-0.0508
2410.02800.04570.0741
242-0.0429-0.0325-0.0665
243-0.0338-0.0532-0.0641
2440.01140.03340.0421
2450.14630.16000.1285
246-0.0275-0.0250-0.0404
247-0.0496-0.0207-0.0420
2480.02960.0326-0.0229
2490.05500.08320.0890
Source: Emolia / Emilia source archive · SHA-256 4e2cf3bc760aeb10…
e03EnglishEmolia / Emilia

Emolia EN · EN_B00000_S00210_W000012

Natural speech · Original speech · 3.4 s

Great episode. Uh, never expected Robin to be fine.

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Hope Enthusiasm Optimism
z +0.61 / +1.10raw 0.58 / 0.96
Elation
z +0.52 / +0.80raw 0.39 / 0.59
Contentment
z +0.48 / +0.61raw 0.25 / 0.32
Concentration
z +0.46 / +0.38raw 0.98 / 0.91
Pleasure Ecstasy
z +0.33 / +0.36raw 0.22 / 0.24
Astonishment Surprise
z +0.05 / +0.27raw 0.03 / 0.16
Pride
z +0.26 / +0.08raw 0.13 / 0.04

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Chest resonance (R_CHST)
z +1.41 / +1.37raw 3.32 / 3.29
Metallic character (METL)
z +1.19 / +1.35raw 2.76 / 2.92
Valence shift (VALS)
z +0.90 / +1.16raw 3.48 / 3.77
Oral resonance (R_ORAL)
z +1.15 / +1.08raw 3.03 / 2.98
Authoritative style (S_AUTH)
z +0.94 / +1.10raw 2.07 / 2.23
Arousal shift (ARSH)
z +0.72 / +1.03raw 3.21 / 3.47
Playful style (S_PLAY)
z +0.66 / +0.88raw 3.22 / 3.51

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z +1.24 / +1.20raw 4.34 / 4.28
Vocal-burst blend (0–10)z -1.61 / -1.70raw 2.27 / 2.01
Content enjoyment (CE)z -0.39 / +0.23raw 4.40 / 5.03
Production quality (PQ)z -1.23 / -0.75raw 5.96 / 6.39
Speech signal quality, calibratedz -0.27 / -0.15raw 3.35 / 3.41
Background quality, calibratedz +0.15 / +0.08raw 3.53 / 3.46
Overall quality, calibratedz -0.03 / +0.00raw 2.82 / 2.84
P.808 MOS estimatez +0.46 / +0.12raw 3.65 / 3.48
score content enjoymentz +0.04 / +0.03raw 4.52 / 4.52
score overall qualityz -0.18 / -0.36raw 2.56 / 2.47
Background Noisez -0.56 / -0.28raw 0.66 / 0.73
Tempo (TEMP)z +0.52 / +0.56raw 2.73 / 2.79
Arousal (AROU)z +0.52 / +0.51raw 2.96 / 2.94
Valence (VALN)z +0.58 / +0.73raw 2.98 / 3.14
Recording quality (RCQL)z -0.34 / -0.24raw 2.45 / 2.54
Background noise (BKGN)z +0.68 / +0.71raw 3.22 / 3.24
talking speedz +2.21 / +1.65raw 17.34 / 15.23
Characters per second (CPS): 14.80 raw · z +1.54

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.6690.667
Identity · 250D0.5640.610

Vocal-burst regions

Tiny
Base
0 s3.4 s

Verified burst intervals are unavailable for this clip.

Tiny: 1 regions, count-head estimate 1.06; Base: 1 regions, count-head estimate 1.19.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
Tiny11.201.600.40
Base11.201.620.42
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.031+0.040.001-0.010.0070.570
1Amusementemo_Amusement0.074+0.080.057+0.07-0.0010.885
2Angeremo_Anger-0.018-0.040.041+0.060.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.032+0.050.157+0.270.0050.572
4Aweemo_Awe0.007+0.020.009+0.020.0000.427
5Bitternessemo_Bitterness0.034+0.050.006-0.020.0130.415
6Concentrationemo_Concentration0.979+0.460.913+0.380.6340.741
7Confusionemo_Confusion0.052+0.090.001-0.000.0010.540
8Contemplationemo_Contemplation0.036-0.080.002-0.130.0820.598
9Contemptemo_Contempt-0.020-0.040.028+0.06-0.0000.464
10Contentmentemo_Contentment0.245+0.480.316+0.610.0000.514
11Disappointmentemo_Disappointment0.068+0.100.057+0.080.0000.700
12Disgustemo_Disgust0.010+0.010.040+0.070.0070.489
13Distressemo_Distress0.044+0.050.037+0.050.0000.821
14Doubtemo_Doubt0.053+0.100.059+0.110.0060.466
15Elationemo_Elation0.390+0.520.592+0.800.0120.728
16Embarrassmentemo_Embarrassment0.117+0.170.130+0.200.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.129+0.140.126+0.130.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion-0.002-0.060.013-0.030.0280.539
19Fearemo_Fear-0.032-0.08-0.049-0.100.0150.610
20Helplessnessemo_Helplessness-0.014-0.02-0.048-0.06-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.578+0.610.957+1.100.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.031-0.170.049-0.140.1490.705
23Infatuationemo_Infatuation0.034+0.090.003+0.03-0.0150.533
24Interestemo_Interest1.532-0.061.671+0.091.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness-0.000+0.04-0.007+0.03-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.054-0.010.019-0.090.0590.442
27Longingemo_Longing0.015-0.11-0.044-0.210.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.067+0.020.038-0.030.0550.534
29Painemo_Pain-0.000+0.00-0.007-0.01-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.221+0.330.237+0.360.0170.614
31Prideemo_Pride0.133+0.260.044+0.080.0020.509
32Reliefemo_Relief0.076+0.060.168+0.200.0330.681
33Sadnessemo_Sadness0.059+0.080.002+0.000.0000.759
34Sexual Lustemo_Sexual_Lust-0.021-0.10-0.025-0.110.0300.494
35Shameemo_Shame0.010-0.01-0.003-0.030.0130.510
36Sournessemo_Sourness0.042+0.110.007+0.03-0.0050.439
37Teasingemo_Teasing0.028+0.050.024+0.040.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.130+0.130.197+0.220.0410.697
39Triumphemo_Triumph0.020+0.05-0.023-0.03-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg2.202-0.632.143-0.682.9841.238
41Arousal (AROU)vn_AROU_reg2.958+0.522.943+0.512.3591.147
42Arousal shift (ARSH)vn_ARSH_reg3.207+0.723.471+1.032.5800.867
43Vocal attack (ATCK)vn_ATCK_reg2.515+0.272.633+0.362.1621.295
44Background noise (BKGN)vn_BKGN_reg3.222+0.683.240+0.712.7030.761
45Brightness (BRGT)vn_BRGT_reg2.272+0.352.277+0.351.9910.813
46Speech chunking (CHNK)vn_CHNK_reg2.491+0.522.330+0.391.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg2.015-0.002.092+0.082.0180.945
48Cognitive load (COGL)vn_COGL_reg2.419+0.312.320+0.212.0941.061
49Dynamic arc (DARC)vn_DARC_reg1.684+0.051.749+0.101.6151.380
50Disfluency (DFLU)vn_DFLU_reg3.156+0.563.073+0.492.4881.184
51Emphasis (EMPH)vn_EMPH_reg2.853+0.163.233+0.412.6071.524
52Esthetics (ESTH)vn_ESTH_reg2.733+0.422.666+0.342.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.306+0.680.263+0.550.0860.322
54Focus (FOCS)vn_FOCS_reg2.391+0.042.384+0.042.3441.102
55Fullness (FULL)vn_FULL_reg2.198+0.472.434+0.741.7780.889
56Perceived gender (GEND)vn_GEND_reg4.939+0.564.852+0.504.0551.588
57Harmonicity (HARM)vn_HARM_reg2.918+0.522.819+0.412.4800.836
58Metallic character (METL)vn_METL_reg2.761+1.192.918+1.351.6210.958
59Pitch range (RANG)vn_RANG_reg2.786-0.142.801-0.132.9140.897
60Recording quality (RCQL)vn_RCQL_reg2.451-0.342.541-0.242.7540.888
61Vocal register (REGS)vn_REGS_reg0.667-0.420.878-0.251.1981.268
62Respiration (RESP)vn_RESP_reg1.510-1.111.376-1.232.6801.056
63Roughness (ROUG)vn_ROUG_reg1.155-0.991.585-0.512.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg3.321+1.413.288+1.372.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg2.613+0.572.825+0.822.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg2.614+0.702.722+0.811.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg2.086+0.772.069+0.751.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg3.126+0.323.095+0.292.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg3.033+1.152.981+1.082.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg2.716+0.802.744+0.842.0700.803
71Smoothness (SMTH)vn_SMTH_reg2.462+0.292.497+0.322.1431.093
72Stance (STNC)vn_STNC_reg2.914+0.253.072+0.392.6331.137
73Speech structure (STRU)vn_STRU_reg2.638+0.522.618+0.502.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg0.025-0.950.268-0.781.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg2.072+0.942.230+1.101.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg0.391-0.830.947-0.441.5601.401
77Casual style (S_CASU)vn_S_CASU_reg4.493+0.554.473+0.533.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg3.635+0.793.589+0.752.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg0.866-1.621.008-1.492.7211.148
80Formal style (S_FORM)vn_S_FORM_reg1.458+0.261.484+0.281.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg2.414-0.322.369-0.352.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg1.038-0.190.998-0.211.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.463+0.100.501+0.160.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg3.218+0.663.508+0.882.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg0.258-0.740.507-0.551.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg1.574-0.661.659-0.602.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg1.014+0.300.971+0.260.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg-0.094-1.05-0.144-1.081.5971.618
89Tempo (TEMP)vn_TEMP_reg2.734+0.522.786+0.562.1171.190
90Tension (TENS)vn_TENS_reg1.687+0.151.686+0.151.5590.852
91Valence (VALN)vn_VALN_reg2.976+0.583.144+0.732.3281.118
92Valence shift (VALS)vn_VALS_reg3.485+0.903.768+1.162.5181.079
93Velocity flux (VFLX)vn_VFLX_reg2.430+0.142.570+0.262.2561.208
94Volatility (VOLT)vn_VOLT_reg2.026-0.632.259-0.372.5860.886
95Vulnerability (VULN)vn_VULN_reg2.627-0.022.538-0.112.6480.984
96Warmth (WARM)vn_WARM_reg2.388-0.212.613+0.042.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_64.338+1.244.275+1.202.3321.621
98Vocal-burst blend (0–10)blend_0_102.270-1.612.015-1.706.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.961-0.400.920-0.521.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.225-0.093.249-0.013.2500.293
101Arousaleiv_extra:Arousal1.121-0.381.356+0.151.2890.435
102Authenticityeiv_extra:Authenticity2.665-0.012.676+0.062.6670.153
103Background Noiseeiv_extra:Background_Noise0.658-0.560.734-0.280.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.051+0.041.122+0.271.0400.303
105Gendereiv_extra:Gender-0.853-0.23-0.713-0.09-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.915+0.351.823-0.021.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive1.378-1.021.655-0.431.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.168+0.072.125-0.062.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous1.089+0.471.227+0.870.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.088-0.600.283+0.010.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.397+0.070.355-0.060.3760.324
112Valenceeiv_extra:Valence-0.002-0.020.227+0.580.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached2.215+1.151.985+0.621.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold0.084+0.110.226+0.580.0510.300
115score background qualityeiv_extra:score_background_quality3.507+0.303.328+0.043.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.519+0.044.515+0.034.5070.303
117score overall qualityeiv_extra:score_overall_quality2.560-0.182.468-0.362.6520.507
118score speech qualityeiv_extra:score_speech_quality2.016+1.431.928+0.831.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE4.398-0.395.034+0.234.7991.018
120Content usefulness (CU)audiobox:CU4.930-0.725.839+0.195.6491.001
121Production complexity (PC)audiobox:PC1.915-0.262.117-0.132.3311.617
122Production quality (PQ)audiobox:PQ5.961-1.236.387-0.757.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw3.680-0.333.774-0.193.9000.669
124Background quality, rawdnsmos:BAK_raw3.456+0.223.368+0.143.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.075-0.013.095+0.013.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.351-0.273.415-0.153.4880.501
127Background quality, calibrateddnsmos:BAK3.528+0.153.464+0.083.3890.928
128Overall quality, calibrateddnsmos:OVRL2.820-0.032.842+0.002.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.647+0.463.480+0.123.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.725+1.120.784+1.210.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.396-0.990.201-1.350.9470.554
132Agevoiceclap_attribute:Age3.126-0.193.081-0.273.2290.547
133Amusementvoiceclap_attribute:Amusement0.287-0.180.408+0.030.3920.572
134Angervoiceclap_attribute:Anger0.545-0.880.292-1.390.9920.506
135Arousalvoiceclap_attribute:Arousal1.253-0.271.295-0.201.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.693+0.380.913+0.840.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.656-0.682.635-0.812.7700.167
138Awevoiceclap_attribute:Awe0.322-0.840.415-0.480.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.375-0.730.479-0.330.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.367-0.960.221-1.570.5920.236
141Concentrationvoiceclap_attribute:Concentration0.967-0.470.868-0.651.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.498+0.921.399+0.701.0690.468
143Confusionvoiceclap_attribute:Confusion0.675+0.100.672+0.080.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.347-1.070.354-1.050.7800.406
145Contemptvoiceclap_attribute:Contempt0.407-0.890.287-1.240.7090.339
146Contentmentvoiceclap_attribute:Contentment0.201-1.260.249-1.100.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.248-1.760.161-2.060.7600.291
148Disgustvoiceclap_attribute:Disgust0.353-1.310.279-1.670.6250.207
149Distressvoiceclap_attribute:Distress0.431-1.420.390-1.530.9270.350
150Doubtvoiceclap_attribute:Doubt0.574+0.210.583+0.250.5350.188
151Elationvoiceclap_attribute:Elation0.506-0.080.671+0.330.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.461+0.490.352+0.140.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.663-0.000.664+0.000.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.434-0.110.298-0.630.4620.261
155Fearvoiceclap_attribute:Fear0.190-2.020.257-1.700.6180.212
156Gendervoiceclap_attribute:Gender-0.794-0.64-0.826-0.67-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.328-1.230.284-1.390.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.947+0.291.918+0.211.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.612-0.240.713+0.020.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability0.722-0.610.443-1.131.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.230-1.140.123-1.460.6080.332
162Interestvoiceclap_attribute:Interest1.327-0.621.360-0.541.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.280-0.480.292-0.450.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.066-1.860.033-2.000.5230.245
165Longingvoiceclap_attribute:Longing0.311-1.030.271-1.210.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.260-1.660.133-2.030.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive1.311-0.781.364-0.691.7750.594
168Painvoiceclap_attribute:Pain0.186-0.900.169-0.980.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.315-0.630.437-0.220.5030.301
170Pridevoiceclap_attribute:Pride0.261-1.490.340-1.260.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.167-0.452.170-0.442.3130.328
172Reliefvoiceclap_attribute:Relief0.621-0.970.688-0.811.0250.415
173Sadnessvoiceclap_attribute:Sadness0.406-1.110.381-1.180.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous1.016+0.371.176+0.650.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.118-1.240.000-1.520.6350.419
176Shamevoiceclap_attribute:Shame0.336-0.670.294-0.820.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.119-0.110.106-0.150.1600.360
178Sournessvoiceclap_attribute:Sourness0.376-1.130.326-1.310.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.309-0.880.233-1.090.6230.359
180Teasingvoiceclap_attribute:Teasing0.423-0.310.383-0.420.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude0.672-1.340.507-1.661.3840.529
182Triumphvoiceclap_attribute:Triumph0.422-0.820.539-0.490.7160.359
183Valencevoiceclap_attribute:Valence-0.347-0.540.276+0.45-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached2.282+1.162.226+1.061.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold0.141+0.520.087+0.38-0.0660.400
186durationvoiceclap_attribute:duration4.807-0.965.153-0.9110.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.388+0.293.306+0.123.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.604-0.354.594-0.424.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.411-0.542.429-0.502.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality2.077+2.682.040+2.151.8910.069
191talking speedvoiceclap_attribute:talking_speed17.338+2.2115.227+1.659.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.08440.0171-0.0802
10.08260.0768-0.0732
2-0.0413-0.0449-0.0533
3-0.01500.0088-0.0858
40.23810.21110.2002
50.0304-0.02610.1136
6-0.0966-0.1254-0.0788
70.06710.12540.1583
8-0.0073-0.0483-0.0745
9-0.0535-0.04420.0908
10-0.0498-0.0523-0.0173
11-0.0686-0.04360.0419
12-0.1045-0.14110.0018
13-0.1233-0.0998-0.0858
14-0.0235-0.04600.0267
150.02050.02000.0271
16-0.0011-0.0111-0.0062
170.02780.07580.1692
180.03750.05280.0576
190.20650.21840.1345
20-0.0966-0.0836-0.0870
21-0.0272-0.0091-0.0621
22-0.1264-0.1239-0.0817
23-0.1276-0.1108-0.1028
240.0302-0.01050.0398
25-0.03830.02340.0443
260.04500.06660.0648
27-0.1069-0.0862-0.0811
28-0.1124-0.0737-0.0950
290.0194-0.03200.1702
300.15920.20380.1405
310.09660.1103-0.0803
32-0.0014-0.03740.0682
330.07350.06380.0869
340.04160.02640.0962
35-0.0790-0.1145-0.0231
36-0.02570.01250.1246
370.00520.0163-0.0581
38-0.0738-0.0337-0.0867
39-0.0178-0.06900.0723
400.0333-0.01090.1660
410.01680.02510.0128
42-0.01770.06560.0770
43-0.1057-0.00660.0237
440.0114-0.0027-0.0046
450.08690.07110.0632
46-0.0556-0.0384-0.0820
47-0.1294-0.1014-0.0341
480.17980.10560.0569
49-0.0954-0.1118-0.0878
500.05470.05040.0456
51-0.0747-0.1275-0.0869
52-0.0796-0.1092-0.0988
53-0.1136-0.1322-0.0718
540.01740.06190.0904
550.0112-0.0157-0.0836
560.05830.06510.0642
570.07320.0554-0.0328
58-0.0802-0.05170.0902
59-0.0960-0.0690-0.0872
600.0477-0.01530.1074
61-0.0556-0.0517-0.0800
62-0.0747-0.0998-0.0813
630.16520.21220.1826
64-0.1075-0.1411-0.0882
65-0.0693-0.0747-0.0709
660.11850.0758-0.1130
67-0.0905-0.11340.0295
68-0.0029-0.00780.0390
69-0.1245-0.0909-0.0797
700.04160.05360.1060
71-0.0378-0.1134-0.0963
72-0.0629-0.0666-0.0462
730.00890.01790.0063
74-0.0623-0.0653-0.0950
750.06470.1254-0.0271
760.18950.23100.2321
770.07590.02550.0142
780.06110.06790.1053
79-0.0702-0.0664-0.0754
800.04920.1024-0.0273
81-0.0372-0.0857-0.0787
82-0.0729-0.1056-0.0832
830.0431-0.01710.0423
84-0.1664-0.1421-0.1078
85-0.1470-0.1536-0.0923
860.0228-0.0199-0.1183
87-0.0483-0.0178-0.0526
880.08990.06170.0361
890.05710.03920.0867
90-0.1251-0.1359-0.0942
910.09230.1192-0.0289
92-0.1233-0.0606-0.0766
930.10020.11650.2362
94-0.03950.0463-0.0241
95-0.1221-0.1192-0.0934
96-0.0863-0.0648-0.0943
970.11060.09720.0340
980.00240.1040-0.0391
990.11300.01170.0207
1000.03630.0593-0.0602
101-0.0802-0.0572-0.0840
102-0.1355-0.1087-0.0930
103-0.1051-0.1150-0.0917
104-0.1306-0.1108-0.0850
105-0.1470-0.0779-0.0838
106-0.0905-0.0293-0.0580
1070.20530.17040.1807
1080.02450.04910.0423
1090.00340.0013-0.0726
1100.0000-0.00330.0901
111-0.0893-0.0962-0.1008
112-0.0680-0.0685-0.0324
1130.04860.05410.0214
114-0.1087-0.1212-0.0936
115-0.0623-0.04960.0451
116-0.0240-0.0530-0.0745
117-0.0984-0.1092-0.1011
118-0.0276-0.0967-0.0482
119-0.0024-0.0144-0.0840
1200.11970.05100.0514
121-0.1148-0.1066-0.0881
122-0.1142-0.0909-0.0893
1230.10080.02840.0792
124-0.04500.0348-0.1132
1250.08930.08830.0254
126-0.1027-0.0826-0.0749
127-0.0076-0.0909-0.0803
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.01070.0127-0.0537
1-0.01170.03350.1185
2-0.0356-0.0611-0.0600
3-0.0739-0.0844-0.0196
40.10080.07470.2224
5-0.0093-0.0795-0.0510
60.0231-0.00400.0681
7-0.0981-0.2029-0.0494
8-0.1319-0.1388-0.0423
90.06220.0371-0.0575
10-0.0228-0.0015-0.0512
110.02000.02840.0437
120.08840.14520.1318
130.08220.01100.0385
140.11120.05630.0246
150.02380.0137-0.0054
160.11600.16230.0782
17-0.1064-0.0462-0.0586
18-0.0266-0.0256-0.0497
190.00690.0200-0.0774
20-0.0142-0.0515-0.0488
21-0.0186-0.0267-0.0452
22-0.0642-0.0121-0.0506
23-0.0021-0.0264-0.0472
240.05280.0726-0.0417
25-0.1485-0.1623-0.0897
26-0.0266-0.0694-0.0538
27-0.0798-0.0398-0.0396
28-0.0687-0.0550-0.0528
290.12570.16980.1732
30-0.00890.0162-0.0424
310.01040.0005-0.0490
320.0018-0.0243-0.0502
33-0.0988-0.0577-0.0438
340.22100.14630.1868
35-0.00060.0212-0.0628
36-0.0206-0.0238-0.0515
37-0.0656-0.0398-0.0337
380.0515-0.0403-0.0477
39-0.00740.06620.0599
40-0.0252-0.03080.0012
410.02300.0101-0.0360
42-0.0275-0.0088-0.0485
430.04730.0408-0.0576
44-0.0323-0.0416-0.0587
45-0.0463-0.0699-0.0560
460.03130.0507-0.0717
47-0.0435-0.0014-0.0638
480.0110-0.0304-0.0521
490.0155-0.0141-0.0104
50-0.00840.0275-0.0005
51-0.1188-0.0467-0.0494
52-0.0306-0.0314-0.0481
530.0036-0.02140.0352
54-0.0371-0.03710.0948
550.07800.05900.1125
560.0235-0.0096-0.0671
570.17270.23920.0934
58-0.0139-0.0406-0.0468
590.05210.00300.0522
600.02140.07050.0697
61-0.0266-0.0398-0.0453
62-0.03830.0080-0.0429
63-0.1367-0.0769-0.0472
640.10080.09880.0298
65-0.0663-0.0108-0.0595
660.0003-0.0232-0.0356
670.06700.0153-0.0169
68-0.0406-0.0627-0.0525
69-0.0092-0.0432-0.0634
70-0.0421-0.0095-0.0406
71-0.03870.01210.0095
720.0235-0.0016-0.0540
730.11120.08920.0956
74-0.0608-0.0606-0.0526
75-0.0268-0.0340-0.0673
760.01150.0168-0.0464
77-0.00040.0376-0.0756
780.16640.09770.0562
79-0.0815-0.0822-0.0522
800.01680.0260-0.0518
81-0.0912-0.0683-0.0584
82-0.0570-0.0790-0.0468
830.01950.00290.0224
84-0.0919-0.0886-0.0734
85-0.0392-0.0630-0.0521
86-0.01730.0406-0.0465
87-0.0142-0.0638-0.0416
880.02230.01390.0252
89-0.04070.0136-0.0471
900.15190.1858-0.0516
910.07530.0758-0.0515
920.0268-0.02590.1105
93-0.0642-0.0897-0.0401
94-0.0780-0.0539-0.0517
95-0.0539-0.0758-0.0544
96-0.0155-0.0336-0.0207
97-0.0618-0.0758-0.0613
980.06320.10460.0472
99-0.01130.05230.0272
1000.05870.0526-0.0107
101-0.1264-0.0678-0.0492
1020.00350.0411-0.0287
1030.07320.03580.0035
104-0.0756-0.0459-0.0679
105-0.0435-0.0483-0.0677
106-0.05250.0017-0.0401
1070.08460.04190.0989
1080.04230.09020.1136
109-0.0335-0.0100-0.0501
1100.01780.03920.0067
1110.02420.04720.1012
112-0.0480-0.0382-0.0421
1130.0660-0.0763-0.0888
114-0.03190.0063-0.0499
1150.08910.10620.0023
1160.0288-0.0113-0.0935
1170.16990.17080.0889
1180.0086-0.0352-0.0486
119-0.0843-0.0387-0.0536
120-0.03830.03550.1401
121-0.0135-0.0028-0.0446
122-0.0245-0.0207-0.0188
123-0.0632-0.0478-0.0538
124-0.0371-0.0120-0.0519
1250.08460.10410.0376
1260.08320.05180.0419
1270.00410.0020-0.0290
128-0.00170.0510-0.0433
1290.0095-0.0162-0.0552
130-0.0090-0.0547-0.0523
131-0.0199-0.0053-0.0456
132-0.0011-0.0304-0.0501
133-0.03210.01090.0177
134-0.0981-0.0710-0.0992
135-0.0309-0.0675-0.0524
1360.0067-0.03710.0213
137-0.00490.0120-0.0351
1380.01140.04000.0953
1390.00920.09080.0291
140-0.0856-0.0646-0.0614
1410.05460.05130.0061
1420.06040.05850.0435
143-0.05010.00390.0436
1440.00100.00690.0210
145-0.0149-0.0555-0.0344
146-0.0354-0.0486-0.0624
1470.01390.03150.0257
1480.03000.0478-0.0574
149-0.0079-0.0219-0.0670
1500.0843-0.0019-0.0512
151-0.0663-0.0382-0.0593
1520.02810.03840.0576
1530.04870.0395-0.0577
154-0.0912-0.1009-0.0508
1550.17060.14090.0428
156-0.0691-0.0204-0.0522
1570.09530.08490.1953
158-0.0237-0.0483-0.0648
159-0.1077-0.0737-0.0610
160-0.0430-0.0795-0.0495
161-0.0960-0.0239-0.0258
1620.09460.08380.1606
1630.01720.0153-0.0410
164-0.0559-0.0817-0.0457
1650.16300.11960.1509
166-0.0104-0.0531-0.0606
167-0.03450.0083-0.0660
1680.06980.0179-0.0736
1690.11330.14950.0538
170-0.0656-0.0483-0.0487
171-0.0171-0.0395-0.0458
172-0.0794-0.0566-0.0603
173-0.0432-0.0196-0.0538
1740.06180.02580.0670
1750.04450.0224-0.0495
1760.07220.14410.1768
177-0.02230.00800.0036
1780.07250.09290.0021
179-0.0437-0.0699-0.0641
180-0.0680-0.0590-0.0596
181-0.0760-0.0432-0.0742
182-0.0594-0.0491-0.0450
183-0.0649-0.03980.0266
184-0.0096-0.0326-0.0529
1850.02880.0665-0.0084
186-0.0525-0.0523-0.0628
1870.02990.0451-0.0592
188-0.0656-0.0606-0.0489
1890.0217-0.0190-0.0757
190-0.1430-0.1538-0.0537
191-0.0116-0.01480.0575
1920.0442-0.0657-0.0630
1930.06180.0254-0.0375
1940.06490.07630.0433
1950.04060.0363-0.0495
1960.07460.0542-0.0319
1970.04770.07630.0488
198-0.0347-0.0627-0.0503
1990.04970.0145-0.0572
200-0.0912-0.0416-0.0602
201-0.0497-0.0121-0.0490
202-0.0732-0.0742-0.1083
203-0.00760.10570.1006
204-0.06150.0849-0.0609
205-0.1209-0.0865-0.0906
206-0.0110-0.0210-0.0051
207-0.0608-0.0382-0.0347
208-0.0556-0.0507-0.0540
2090.00250.01310.0157
210-0.0166-0.0833-0.0677
211-0.00140.03040.0171
212-0.0483-0.0913-0.0493
2130.08250.0215-0.0682
2140.0919-0.01370.0417
215-0.0791-0.0031-0.0351
216-0.0542-0.1255-0.0623
2170.01710.09340.0715
218-0.0470-0.0472-0.0604
219-0.0221-0.02750.0409
220-0.05970.0012-0.0575
2210.04040.03260.0036
222-0.0445-0.0156-0.0610
2230.05700.0486-0.0051
2240.0122-0.0360-0.0475
2250.00040.0039-0.0433
226-0.0622-0.0338-0.0841
2270.01110.1057-0.0099
228-0.1361-0.0587-0.0843
2290.01550.0505-0.0571
230-0.0125-0.0087-0.0133
231-0.00990.0085-0.0126
232-0.0174-0.0087-0.0343
233-0.0402-0.0427-0.0528
2340.05420.02230.0149
235-0.0147-0.0472-0.0428
236-0.0774-0.0521-0.0447
237-0.01930.0168-0.0481
2380.01900.01580.0190
239-0.0173-0.0619-0.0538
240-0.0387-0.0545-0.0531
241-0.0656-0.01370.0349
2420.0017-0.0025-0.0269
2430.0245-0.0502-0.0062
244-0.02490.06620.0043
2450.0539-0.0203-0.0495
2460.03180.00190.0953
247-0.0570-0.0218-0.0438
248-0.01040.00660.1053
249-0.0409-0.0571-0.0921
Source: Emolia / Emilia source archive · SHA-256 f8620d4ec74feea8…
e04EnglishEmolia / Emilia

Emolia EN · EN_B00000_S00280_W000023

Natural speech · Original speech · 7.3 s

Very perverted character and a few too many breast references that make it feel a little puerile but it still hits most of the marks.

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Concentration
z +0.90 / +1.09raw 1.30 / 1.44
Contemplation
z +0.29 / +0.93raw 0.26 / 0.64
Contentment
z +0.74 / +0.25raw 0.38 / 0.13
Teasing
z +0.73 / +0.24raw 0.40 / 0.13
Emotional Numbness
z +0.28 / +0.59raw 0.26 / 0.54
Hope Enthusiasm Optimism
z +0.47 / -0.03raw 0.48 / 0.09
Interest
z +0.46 / +0.39raw 2.00 / 1.95

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Newsreader style (S_NEWS)
z +2.63 / +3.10raw 1.90 / 2.17
Fullness (FULL)
z +2.10 / +2.25raw 3.64 / 3.78
Authoritative style (S_AUTH)
z +1.83 / +2.06raw 3.00 / 3.23
Formal style (S_FORM)
z +1.96 / +1.79raw 3.20 / 3.03
Mixed resonance (R_MIXD)
z +1.84 / +1.95raw 2.98 / 3.08
Speech structure (STRU)
z +1.66 / +1.90raw 3.73 / 3.95
Background noise (BKGN)
z +1.68 / +1.78raw 3.98 / 4.06

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z -1.04 / -1.07raw 0.65 / 0.59
Vocal-burst blend (0–10)z -1.59 / -1.73raw 2.31 / 1.93
Content enjoyment (CE)z +1.54 / +1.51raw 6.37 / 6.34
Production quality (PQ)z +0.72 / +0.60raw 7.68 / 7.57
Speech signal quality, calibratedz +0.37 / +0.20raw 3.68 / 3.59
Background quality, calibratedz +0.42 / +0.51raw 3.78 / 3.86
Overall quality, calibratedz +0.66 / +0.68raw 3.24 / 3.25
P.808 MOS estimatez +1.22 / +1.22raw 4.03 / 4.03
score content enjoymentz +1.26 / +1.26raw 4.89 / 4.89
score overall qualityz +0.54 / +0.58raw 2.92 / 2.94
Background Noisez -1.92 / -2.60raw 0.30 / 0.12
Tempo (TEMP)z +0.83 / +1.09raw 3.11 / 3.42
Arousal (AROU)z +0.27 / +0.30raw 2.66 / 2.70
Valence (VALN)z +0.39 / +0.50raw 2.77 / 2.89
Recording quality (RCQL)z +1.36 / +1.36raw 3.96 / 3.96
Background noise (BKGN)z +1.68 / +1.78raw 3.98 / 4.06
talking speedz +2.23 / +2.82raw 17.40 / 19.62
Characters per second (CPS): 18.20 raw · z +2.44

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8160.758
Identity · 250D0.4370.465

Vocal-burst regions

Tiny
Base
0 s7.3 s

Verified burst intervals are unavailable for this clip.

Tiny: 0 regions, count-head estimate 0.00; Base: 0 regions, count-head estimate 0.00.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.134+0.22-0.001-0.010.0070.570
1Amusementemo_Amusement0.305+0.350.098+0.11-0.0010.885
2Angeremo_Anger0.025+0.03-0.002-0.010.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.074+0.120.028+0.040.0050.572
4Aweemo_Awe0.024+0.060.012+0.030.0000.427
5Bitternessemo_Bitterness0.014+0.000.016+0.010.0130.415
6Concentrationemo_Concentration1.304+0.901.441+1.090.6340.741
7Confusionemo_Confusion0.041+0.070.058+0.110.0010.540
8Contemplationemo_Contemplation0.256+0.290.637+0.930.0820.598
9Contemptemo_Contempt0.036+0.080.027+0.06-0.0000.464
10Contentmentemo_Contentment0.379+0.740.131+0.250.0000.514
11Disappointmentemo_Disappointment0.090+0.13-0.003-0.000.0000.700
12Disgustemo_Disgust0.026+0.040.007-0.000.0070.489
13Distressemo_Distress0.065+0.080.009+0.010.0000.821
14Doubtemo_Doubt0.034+0.060.154+0.320.0060.466
15Elationemo_Elation0.270+0.350.062+0.070.0120.728
16Embarrassmentemo_Embarrassment-0.013-0.13-0.039-0.190.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.265+0.280.543+0.590.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.030+0.000.002-0.050.0280.539
19Fearemo_Fear0.046+0.050.061+0.080.0150.610
20Helplessnessemo_Helplessness0.008+0.020.013+0.02-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.477+0.470.092-0.030.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.006-0.200.025-0.180.1490.705
23Infatuationemo_Infatuation-0.013+0.00-0.024-0.02-0.0150.533
24Interestemo_Interest2.003+0.461.945+0.391.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness-0.047-0.04-0.045-0.04-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.041-0.040.072+0.030.0590.442
27Longingemo_Longing-0.002-0.140.005-0.130.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.036-0.040.024-0.060.0550.534
29Painemo_Pain0.032+0.06-0.004-0.00-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.196+0.290.016-0.000.0170.614
31Prideemo_Pride0.006+0.010.017+0.030.0020.509
32Reliefemo_Relief0.051+0.03-0.017-0.070.0330.681
33Sadnessemo_Sadness0.056+0.070.002+0.000.0000.759
34Sexual Lustemo_Sexual_Lust0.145+0.23-0.029-0.120.0300.494
35Shameemo_Shame0.010-0.010.006-0.010.0130.510
36Sournessemo_Sourness0.018+0.050.023+0.06-0.0050.439
37Teasingemo_Teasing0.399+0.730.131+0.240.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.075+0.050.052+0.010.0410.697
39Triumphemo_Triumph0.000+0.01-0.011-0.01-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg3.469+0.393.292+0.252.9841.238
41Arousal (AROU)vn_AROU_reg2.664+0.272.702+0.302.3591.147
42Arousal shift (ARSH)vn_ARSH_reg3.190+0.703.180+0.692.5800.867
43Vocal attack (ATCK)vn_ATCK_reg2.461+0.232.376+0.172.1621.295
44Background noise (BKGN)vn_BKGN_reg3.982+1.684.059+1.782.7030.761
45Brightness (BRGT)vn_BRGT_reg2.185+0.242.382+0.481.9910.813
46Speech chunking (CHNK)vn_CHNK_reg3.586+1.403.820+1.591.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg3.493+1.563.441+1.512.0180.945
48Cognitive load (COGL)vn_COGL_reg0.934-1.091.025-1.012.0941.061
49Dynamic arc (DARC)vn_DARC_reg1.794+0.131.701+0.061.6151.380
50Disfluency (DFLU)vn_DFLU_reg1.209-1.081.043-1.222.4881.184
51Emphasis (EMPH)vn_EMPH_reg2.902+0.192.804+0.132.6071.524
52Esthetics (ESTH)vn_ESTH_reg3.749+1.723.499+1.402.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.607+1.620.554+1.450.0860.322
54Focus (FOCS)vn_FOCS_reg3.672+1.213.741+1.272.3441.102
55Fullness (FULL)vn_FULL_reg3.642+2.103.780+2.251.7780.889
56Perceived gender (GEND)vn_GEND_reg4.721+0.424.883+0.524.0551.588
57Harmonicity (HARM)vn_HARM_reg2.407-0.092.811+0.402.4800.836
58Metallic character (METL)vn_METL_reg1.538-0.091.804+0.191.6210.958
59Pitch range (RANG)vn_RANG_reg3.273+0.403.316+0.452.9140.897
60Recording quality (RCQL)vn_RCQL_reg3.961+1.363.961+1.362.7540.888
61Vocal register (REGS)vn_REGS_reg0.785-0.330.893-0.241.1981.268
62Respiration (RESP)vn_RESP_reg1.501-1.121.384-1.232.6801.056
63Roughness (ROUG)vn_ROUG_reg1.454-0.651.437-0.672.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg3.268+1.343.275+1.352.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg2.552+0.492.684+0.652.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg2.770+0.862.866+0.951.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg2.985+1.843.076+1.951.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg3.079+0.273.172+0.362.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg2.577+0.532.661+0.642.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg2.552+0.602.537+0.582.0700.803
71Smoothness (SMTH)vn_SMTH_reg3.791+1.513.678+1.402.1431.093
72Stance (STNC)vn_STNC_reg2.957+0.282.997+0.322.6331.137
73Speech structure (STRU)vn_STRU_reg3.732+1.663.955+1.902.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg0.898-0.330.758-0.431.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg2.997+1.833.232+2.061.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg0.749-0.580.578-0.701.5601.401
77Casual style (S_CASU)vn_S_CASU_reg3.083-0.563.240-0.433.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg3.298+0.533.441+0.642.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg3.093+0.323.104+0.332.7211.148
80Formal style (S_FORM)vn_S_FORM_reg3.199+1.963.026+1.791.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg2.938+0.092.786-0.032.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg3.729+1.673.661+1.631.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg1.899+2.632.166+3.100.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg2.547+0.152.596+0.192.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg1.003-0.160.978-0.181.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg3.643+0.853.431+0.702.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg1.135+0.401.490+0.690.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg2.710+0.692.585+0.611.5971.618
89Tempo (TEMP)vn_TEMP_reg3.109+0.833.419+1.092.1171.190
90Tension (TENS)vn_TENS_reg1.762+0.241.802+0.291.5590.852
91Valence (VALN)vn_VALN_reg2.769+0.392.888+0.502.3281.118
92Valence shift (VALS)vn_VALS_reg2.747+0.212.916+0.372.5181.079
93Velocity flux (VFLX)vn_VFLX_reg2.876+0.513.093+0.692.2561.208
94Volatility (VOLT)vn_VOLT_reg2.009-0.652.169-0.472.5860.886
95Vulnerability (VULN)vn_VULN_reg2.309-0.352.090-0.572.6480.984
96Warmth (WARM)vn_WARM_reg3.595+1.123.422+0.932.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_60.654-1.040.591-1.072.3321.621
98Vocal-burst blend (0–10)blend_0_102.312-1.591.930-1.736.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.355-2.190.314-2.311.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.253+0.013.339+0.303.2500.293
101Arousaleiv_extra:Arousal1.203-0.201.019-0.621.2890.435
102Authenticityeiv_extra:Authenticity2.807+0.912.774+0.702.6670.153
103Background Noiseeiv_extra:Background_Noise0.301-1.920.121-2.600.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant0.952-0.291.059+0.061.0400.303
105Gendereiv_extra:Gender-0.795-0.17-0.853-0.23-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.935+0.431.969+0.561.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive1.772-0.191.632-0.481.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.707+1.702.738+1.792.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous1.147+0.641.086+0.460.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.287+0.020.191-0.280.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.325-0.160.194-0.560.3760.324
112Valenceeiv_extra:Valence0.116+0.29-0.024-0.080.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached2.433+1.642.440+1.661.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold0.206+0.520.140+0.300.0510.300
115score background qualityeiv_extra:score_background_quality3.496+0.283.524+0.323.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.890+1.264.888+1.264.5070.303
117score overall qualityeiv_extra:score_overall_quality2.923+0.542.945+0.582.6520.507
118score speech qualityeiv_extra:score_speech_quality1.901+0.641.950+0.981.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE6.372+1.546.340+1.514.7991.018
120Content usefulness (CU)audiobox:CU7.236+1.597.244+1.595.6491.001
121Production complexity (PC)audiobox:PC2.041-0.181.820-0.322.3311.617
122Production quality (PQ)audiobox:PQ7.679+0.727.572+0.607.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw4.238+0.514.097+0.293.9000.669
124Background quality, rawdnsmos:BAK_raw3.907+0.624.034+0.733.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.726+0.793.742+0.813.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.675+0.373.587+0.203.4880.501
127Background quality, calibrateddnsmos:BAK3.781+0.423.865+0.513.3890.928
128Overall quality, calibrateddnsmos:OVRL3.243+0.663.250+0.682.8400.606
129P.808 MOS estimatednsmos:P808_MOS4.027+1.224.027+1.223.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.002+0.00-0.008-0.010.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.773-0.310.485-0.830.9470.554
132Agevoiceclap_attribute:Age3.187-0.083.199-0.053.2290.547
133Amusementvoiceclap_attribute:Amusement0.461+0.120.443+0.090.3920.572
134Angervoiceclap_attribute:Anger0.529-0.920.702-0.570.9920.506
135Arousalvoiceclap_attribute:Arousal1.294-0.201.452+0.051.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.383-0.260.550+0.090.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.820+0.302.852+0.492.7700.167
138Awevoiceclap_attribute:Awe0.485-0.210.520-0.070.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.210-1.360.226-1.290.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.270-1.370.305-1.220.5920.236
141Concentrationvoiceclap_attribute:Concentration1.250+0.071.230+0.031.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant0.833-0.500.866-0.431.0690.468
143Confusionvoiceclap_attribute:Confusion0.022-1.860.058-1.750.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.392-0.960.436-0.850.7800.406
145Contemptvoiceclap_attribute:Contempt0.490-0.650.474-0.690.7090.339
146Contentmentvoiceclap_attribute:Contentment0.497-0.300.503-0.280.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.370-1.340.498-0.900.7600.291
148Disgustvoiceclap_attribute:Disgust0.412-1.030.437-0.900.6250.207
149Distressvoiceclap_attribute:Distress0.352-1.640.382-1.560.9270.350
150Doubtvoiceclap_attribute:Doubt0.154-2.020.199-1.790.5350.188
151Elationvoiceclap_attribute:Elation0.626+0.220.672+0.330.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.149-0.500.162-0.460.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.671+0.030.704+0.150.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion-0.002-1.780.081-1.460.4620.261
155Fearvoiceclap_attribute:Fear0.471-0.690.381-1.120.6180.212
156Gendervoiceclap_attribute:Gender-0.532-0.38-0.716-0.56-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.145-1.910.223-1.620.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.979+0.371.938+0.261.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.849+0.361.013+0.780.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability0.349-1.300.618-0.801.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.501-0.320.584-0.070.6080.332
162Interestvoiceclap_attribute:Interest2.089+1.121.983+0.871.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.017-1.050.115-0.840.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.277-1.000.407-0.470.5230.245
165Longingvoiceclap_attribute:Longing0.252-1.300.287-1.140.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.583-0.720.535-0.860.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive1.659-0.201.670-0.181.7750.594
168Painvoiceclap_attribute:Pain0.146-1.080.186-0.900.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.376-0.420.471-0.110.5030.301
170Pridevoiceclap_attribute:Pride0.427-1.010.476-0.870.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.912+1.832.887+1.752.3130.328
172Reliefvoiceclap_attribute:Relief0.559-1.120.691-0.801.0250.415
173Sadnessvoiceclap_attribute:Sadness0.420-1.060.447-0.970.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.909+0.190.791-0.020.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.474-0.380.564-0.170.6350.419
176Shamevoiceclap_attribute:Shame0.075-1.600.198-1.160.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.283+0.340.144-0.040.1600.360
178Sournessvoiceclap_attribute:Sourness0.390-1.090.545-0.550.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.499-0.350.628+0.010.6230.359
180Teasingvoiceclap_attribute:Teasing0.612+0.190.640+0.260.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.096-0.541.299-0.161.3840.529
182Triumphvoiceclap_attribute:Triumph0.356-1.000.375-0.950.7160.359
183Valencevoiceclap_attribute:Valence0.434+0.690.281+0.45-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached2.430+1.422.295+1.181.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold0.371+1.090.241+0.77-0.0660.400
186durationvoiceclap_attribute:duration8.271-0.427.528-0.5310.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.517+0.553.599+0.713.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment5.011+2.425.048+2.674.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.884+0.603.029+0.952.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.960+0.991.983+1.331.8910.069
191talking speedvoiceclap_attribute:talking_speed17.397+2.2319.624+2.829.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.12230.14840.1231
1-0.1168-0.0711-0.1025
20.04710.06350.0450
3-0.1168-0.1246-0.1046
40.19680.20150.1531
5-0.00160.0220-0.0231
60.06390.05400.0811
7-0.0612-0.0873-0.0945
8-0.0475-0.05990.0787
90.06700.08000.0744
10-0.0724-0.07880.0042
11-0.1339-0.0947-0.0468
120.03470.04490.0389
13-0.0950-0.0983-0.1019
140.15990.22230.1390
15-0.0646-0.0635-0.0925
16-0.0553-0.0721-0.0378
170.09220.08370.1158
180.04850.07020.0536
19-0.0594-0.0812-0.0687
20-0.1360-0.1331-0.1033
210.09570.05220.1526
22-0.0458-0.0231-0.0742
23-0.1503-0.1356-0.0894
24-0.0724-0.0824-0.0833
25-0.0871-0.0788-0.0906
260.10040.0660-0.0118
27-0.0420-0.0534-0.0367
28-0.0991-0.1087-0.0085
290.21450.21620.1782
300.17630.19420.0881
310.08540.06410.1543
320.19400.12280.1316
330.07580.11300.0623
34-0.0401-0.00260.0047
350.0004-0.0278-0.1118
360.15310.18080.1578
37-0.0169-0.0064-0.0058
38-0.0319-0.0038-0.0425
39-0.01210.0257-0.0062
40-0.0635-0.0653-0.0304
41-0.00400.0773-0.0086
420.0337-0.05040.0765
43-0.1004-0.0959-0.1062
440.01350.00520.1060
45-0.0214-0.0313-0.0399
460.04320.02950.0884
47-0.0635-0.0681-0.0657
48-0.0362-0.0182-0.0560
49-0.0888-0.0766-0.0662
500.05910.01110.0638
51-0.1209-0.1093-0.1031
52-0.02390.02260.0865
53-0.0902-0.0989-0.1052
540.12090.09340.0766
55-0.0991-0.1075-0.0993
560.05400.07390.1753
57-0.0112-0.0051-0.0487
58-0.0998-0.1124-0.1004
59-0.0085-0.06500.0335
600.03810.0886-0.0854
61-0.0765-0.0867-0.0820
62-0.0290-0.00510.0248
630.02440.0269-0.0351
64-0.01530.0155-0.0445
650.04780.03850.0259
660.19400.16000.1332
670.09770.09280.1042
680.06800.05650.0168
69-0.1093-0.0788-0.0945
70-0.0625-0.0632-0.1069
710.01610.01440.0029
72-0.02240.0209-0.0770
730.12850.14170.0031
74-0.0697-0.0892-0.0426
750.21590.22960.1823
76-0.0196-0.0574-0.1001
77-0.0741-0.0873-0.0792
78-0.0225-0.0108-0.0404
79-0.1073-0.1069-0.0895
800.26370.25650.2336
81-0.0929-0.0934-0.0935
820.02190.0776-0.0148
830.00580.0034-0.0499
84-0.03570.00670.0299
85-0.00040.00060.0640
860.01920.02640.0117
87-0.0408-0.07450.0146
880.07240.03440.0866
890.07380.03650.0427
900.03070.1160-0.0643
910.05260.00920.0061
92-0.0868-0.0760-0.0671
93-0.0871-0.0837-0.1008
94-0.0139-0.0065-0.0722
95-0.0478-0.0068-0.0285
96-0.0902-0.1301-0.0983
97-0.1209-0.0818-0.1103
980.0006-0.07760.0803
99-0.0425-0.0176-0.0956
1000.14140.09470.1138
1010.02200.03680.0594
102-0.0362-0.03600.0890
103-0.1155-0.0782-0.0693
1040.02530.06720.1016
105-0.0902-0.1038-0.0546
106-0.0799-0.0782-0.0628
1070.12910.09220.0708
1080.11070.07390.1138
109-0.0963-0.1160-0.1013
1100.0016-0.03630.1373
111-0.0378-0.0431-0.0713
112-0.0553-0.0208-0.1017
1130.02320.0452-0.0028
114-0.0430-0.0403-0.0054
115-0.1052-0.0730-0.0978
116-0.0868-0.0550-0.0884
117-0.0506-0.09410.0470
1180.01240.01680.0604
1190.03310.03420.0630
1200.13460.06410.0085
121-0.1462-0.1515-0.1046
1220.07000.07600.2295
123-0.0758-0.0849-0.1050
1240.08270.10080.0975
125-0.0301-0.0051-0.0768
126-0.0687-0.0910-0.0889
1270.02370.0020-0.0438
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.0811-0.05900.0321
10.0058-0.0173-0.0482
20.01770.10340.1159
3-0.0551-0.02900.0413
4-0.0411-0.0583-0.0453
5-0.01270.00960.0475
6-0.0015-0.03960.0273
70.15050.13700.0511
80.0048-0.01430.0142
9-0.0007-0.0335-0.0820
10-0.0338-0.06910.0292
11-0.0326-0.0517-0.0674
120.03340.0193-0.0039
13-0.0776-0.0175-0.0440
14-0.0124-0.0182-0.0712
150.08150.04310.0042
160.06790.1763-0.0031
170.0299-0.03120.0538
180.05470.0431-0.0508
19-0.0729-0.0243-0.0791
20-0.04020.01760.1155
21-0.0303-0.0361-0.0462
22-0.01780.0047-0.0517
230.10940.09960.0019
24-0.0312-0.0377-0.0426
25-0.0012-0.02580.1543
26-0.06280.0078-0.0522
27-0.0378-0.0073-0.0405
28-0.04090.00930.0356
29-0.0146-0.0596-0.0480
300.03920.0265-0.0434
31-0.0566-0.0069-0.0626
320.10320.1560-0.0149
33-0.0293-0.0568-0.0345
34-0.01220.0230-0.0243
350.02910.0104-0.0642
36-0.0838-0.0469-0.0527
37-0.0462-0.0732-0.0345
38-0.1117-0.10020.0083
39-0.0644-0.0843-0.0642
40-0.0378-0.0691-0.0577
41-0.1055-0.1459-0.0573
420.0148-0.0403-0.0495
430.05390.03850.0524
440.03340.05360.1417
45-0.0308-0.0691-0.0572
46-0.0158-0.0128-0.0631
470.05390.01020.0603
48-0.0574-0.0767-0.0407
490.06360.04880.0292
50-0.1272-0.0726-0.0533
51-0.1210-0.1560-0.0505
520.00740.01890.0142
530.08500.05490.1116
540.0070-0.0523-0.0190
550.04250.04600.0020
560.06320.0631-0.0355
570.12880.0517-0.0056
58-0.0857-0.0634-0.0193
59-0.0134-0.0523-0.0694
600.0005-0.0258-0.0341
610.00210.0059-0.0352
62-0.1428-0.1446-0.0438
63-0.0857-0.0634-0.0542
64-0.00460.03120.0372
65-0.0683-0.08620.0305
66-0.0671-0.05200.0226
67-0.0687-0.1116-0.0524
68-0.0423-0.03650.0970
690.07290.03200.0377
70-0.0166-0.04060.0304
71-0.0072-0.0396-0.0254
72-0.00320.0157-0.0552
73-0.0328-0.0209-0.0825
74-0.0481-0.0482-0.0437
750.06050.0404-0.0174
760.02970.03220.0078
770.00360.0154-0.0800
780.06520.1478-0.0352
79-0.03880.0121-0.0436
80-0.00490.0025-0.0529
81-0.0555-0.0536-0.0597
82-0.0815-0.0796-0.0479
83-0.03800.0159-0.0443
840.01220.03660.0151
85-0.0497-0.0989-0.0533
860.07600.0767-0.0323
870.14430.13700.0082
88-0.1521-0.12300.0282
89-0.0035-0.0344-0.0226
900.05660.0672-0.0368
910.09930.08120.0891
920.0056-0.0166-0.0460
93-0.0089-0.01670.0638
94-0.0516-0.0694-0.0120
95-0.0586-0.0016-0.0372
96-0.1272-0.0631-0.0476
97-0.00970.01250.0746
980.0811-0.01400.0862
99-0.0421-0.0463-0.0666
1000.00550.03080.0396
1010.00500.0279-0.0498
102-0.0421-0.04150.0716
103-0.0170-0.0520-0.0880
104-0.0543-0.0751-0.0693
1050.03300.0089-0.0232
106-0.0266-0.0336-0.0410
107-0.0749-0.0209-0.0477
108-0.0780-0.0463-0.0396
1090.04810.04190.1777
110-0.0687-0.0777-0.0634
1110.0927-0.0016-0.0439
112-0.0850-0.0669-0.0430
113-0.0289-0.01370.0474
1140.07800.01310.1228
115-0.0578-0.0146-0.0377
1160.05700.0342-0.0554
1170.00860.0009-0.0869
1180.11250.19020.1954
119-0.0520-0.05800.0062
120-0.0663-0.0542-0.0009
1210.01060.0073-0.0517
1220.0378-0.0055-0.0503
123-0.0452-0.0357-0.0550
124-0.0574-0.0682-0.0531
1250.11870.07130.0136
1260.06560.0472-0.0236
127-0.0227-0.02380.0553
128-0.0500-0.0472-0.0663
129-0.0737-0.0596-0.0565
130-0.02240.01720.0143
131-0.0760-0.0729-0.0466
132-0.0252-0.01300.0014
133-0.0547-0.0523-0.0526
1340.13970.0660-0.0184
135-0.0877-0.1015-0.0536
136-0.0725-0.0197-0.0570
1370.0129-0.03170.0205
138-0.0508-0.0732-0.0469
1390.05470.10150.0947
140-0.0663-0.07800.0223
1410.09080.09450.1207
1420.05350.0751-0.0451
143-0.00330.01590.0624
144-0.0756-0.04280.1360
145-0.0807-0.0347-0.0068
1460.05740.02960.0061
1470.0225-0.0047-0.0473
1480.00380.0373-0.0204
1490.03780.1053-0.0222
150-0.00890.0296-0.0376
1510.0702-0.0179-0.0118
1520.00680.0090-0.0590
1530.04690.05490.1440
154-0.0756-0.0843-0.0519
1550.0095-0.0482-0.0695
1560.02170.03500.0384
1570.14740.16170.1428
158-0.0675-0.0146-0.0663
159-0.0046-0.0182-0.0623
160-0.0252-0.0479-0.0205
161-0.0438-0.05580.0078
1620.00930.0174-0.0618
1630.0174-0.00850.0690
1640.00490.06690.1593
1650.0497-0.0036-0.0762
1660.07560.04760.0431
1670.04330.0155-0.0615
168-0.0702-0.0612-0.0650
169-0.1110-0.1306-0.0607
170-0.0238-0.0083-0.0498
171-0.0970-0.0713-0.0468
1720.0092-0.02030.0663
1730.11790.13130.0116
1740.01640.0431-0.0547
175-0.0328-0.0095-0.0167
1760.0297-0.0069-0.0452
177-0.0582-0.0812-0.0439
178-0.01520.00260.0648
1790.02090.01700.0755
1800.0264-0.0193-0.0289
1810.05740.01820.0506
1820.06520.06560.2045
183-0.0861-0.0843-0.1013
1840.03360.0602-0.0324
1850.02470.0168-0.0637
1860.0047-0.0202-0.0004
1870.0388-0.00470.0836
188-0.0053-0.0290-0.0500
189-0.0448-0.0729-0.0153
1900.06630.06340.0361
191-0.0857-0.0457-0.0711
192-0.01210.0453-0.0087
1930.14120.13570.1163
1940.02680.04820.0244
1950.12490.1452-0.0220
196-0.03360.02350.0077
1970.02560.0311-0.0020
198-0.0900-0.0628-0.0514
1990.03740.0173-0.0112
200-0.0341-0.0555-0.0492
2010.0121-0.0422-0.0501
202-0.0382-0.0222-0.0024
203-0.01270.00130.0596
2040.0057-0.04410.0439
205-0.1094-0.1300-0.0625
206-0.01530.0271-0.0068
2070.01720.0214-0.0504
208-0.06170.0101-0.0381
209-0.0151-0.05990.0418
2100.01770.01730.0125
211-0.0493-0.0498-0.0409
2120.08460.10400.0050
2130.19710.15920.1464
214-0.02380.0076-0.0671
2150.16060.12870.1797
2160.10320.05580.1343
2170.06250.04880.1066
2180.03260.06340.1395
219-0.0485-0.0514-0.0197
220-0.0535-0.0498-0.0587
221-0.0539-0.02380.0587
2220.05550.0174-0.0231
2230.01300.02710.0140
224-0.0016-0.01580.0228
225-0.0264-0.0085-0.0216
2260.07290.03920.0808
2270.01730.03870.0735
228-0.02970.01710.0338
2290.04870.07420.1575
2300.01010.07480.0144
231-0.0962-0.0539-0.0736
232-0.0706-0.03090.0033
233-0.03140.01620.0914
2340.03670.02050.0096
2350.12030.18260.0513
236-0.0520-0.0517-0.0457
237-0.0419-0.02120.0645
2380.00950.00520.0873
2390.08220.1027-0.0073
240-0.0950-0.0510-0.0500
241-0.0691-0.1002-0.0718
242-0.01590.02620.0206
243-0.04600.00490.0626
2440.08110.10080.1230
2450.09040.1059-0.0506
246-0.1474-0.1459-0.0431
247-0.05590.00550.0228
2480.03880.03790.0276
2490.04110.0243-0.0277
Source: Emolia / Emilia source archive · SHA-256 506ea3896dd448ab…
e05FrenchEmolia / Emilia

Emolia FR · FR_B00000_S00110_W000008

Natural speech · Original speech · 7.3 s

J'en déduis que son auteur est sans doute assez fortuné car le papier est d'une rare qualité. Oui.

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Contemplation
z +0.97 / +1.29raw 0.66 / 0.85
Doubt
z +1.21 / +0.83raw 0.57 / 0.39
Concentration
z +1.18 / +0.89raw 1.51 / 1.30
Contempt
z +0.31 / +1.05raw 0.14 / 0.49
Sourness
z +0.14 / +0.66raw 0.06 / 0.28
Teasing
z +0.27 / +0.60raw 0.15 / 0.33
Emotional Numbness
z +0.60 / +0.36raw 0.55 / 0.34

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Metallic character (METL)
z +1.39 / +1.39raw 2.96 / 2.96
Throat resonance (R_THRT)
z +1.21 / +1.24raw 3.04 / 3.07
Teacher or didactic style (S_TECH)
z +1.16 / +1.01raw 2.07 / 1.88
Chest resonance (R_CHST)
z +1.04 / +1.08raw 3.01 / 3.05
Conversational style (S_CONV)
z +0.98 / +1.02raw 3.88 / 3.94
Mixed resonance (R_MIXD)
z +0.93 / +1.01raw 2.21 / 2.29
Narrator style (S_NARR)
z +1.00 / +0.69raw 2.75 / 2.30

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z +0.05 / +0.14raw 2.41 / 2.56
Vocal-burst blend (0–10)z -1.60 / -1.75raw 2.29 / 1.89
Content enjoyment (CE)z +0.29 / +0.68raw 5.10 / 5.49
Production quality (PQ)z +0.19 / +0.12raw 7.21 / 7.15
Speech signal quality, calibratedz +0.26 / +0.28raw 3.62 / 3.63
Background quality, calibratedz +0.37 / +0.55raw 3.73 / 3.90
Overall quality, calibratedz +0.53 / +0.70raw 3.16 / 3.26
P.808 MOS estimatez +1.34 / +1.28raw 4.09 / 4.05
score content enjoymentz +0.45 / +0.57raw 4.64 / 4.68
score overall qualityz +0.73 / +0.79raw 3.02 / 3.05
Background Noisez -1.04 / -1.59raw 0.53 / 0.39
Tempo (TEMP)z -0.44 / -0.57raw 1.60 / 1.44
Arousal (AROU)z -0.47 / -0.61raw 1.82 / 1.66
Valence (VALN)z +0.47 / +0.60raw 2.85 / 3.00
Recording quality (RCQL)z -0.48 / -0.50raw 2.33 / 2.31
Background noise (BKGN)z +0.23 / +0.23raw 2.88 / 2.88
talking speedz +0.61 / +0.57raw 11.34 / 11.16
Characters per second (CPS): 13.42 raw · z +1.17

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8020.834
Identity · 250D0.7230.739

Vocal-burst regions

Tiny
Base
0 s7.3 s

Verified burst intervals are unavailable for this clip.

Tiny: 0 regions, count-head estimate 0.69; Base: 1 regions, count-head estimate 0.46.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
Base11.101.500.40
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.011+0.010.009+0.000.0070.570
1Amusementemo_Amusement-0.203-0.23-0.092-0.10-0.0010.885
2Angeremo_Anger0.190+0.300.234+0.370.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.006+0.000.019+0.020.0050.572
4Aweemo_Awe0.014+0.030.027+0.060.0000.427
5Bitternessemo_Bitterness-0.005-0.040.049+0.090.0130.415
6Concentrationemo_Concentration1.510+1.181.296+0.890.6340.741
7Confusionemo_Confusion0.244+0.450.039+0.070.0010.540
8Contemplationemo_Contemplation0.664+0.970.851+1.290.0820.598
9Contemptemo_Contempt0.144+0.310.485+1.05-0.0000.464
10Contentmentemo_Contentment0.059+0.110.176+0.340.0000.514
11Disappointmentemo_Disappointment-0.017-0.02-0.003-0.000.0000.700
12Disgustemo_Disgust-0.001-0.020.111+0.210.0070.489
13Distressemo_Distress0.002+0.00-0.007-0.010.0000.821
14Doubtemo_Doubt0.569+1.210.390+0.830.0060.466
15Elationemo_Elation0.017+0.010.000-0.020.0120.728
16Embarrassmentemo_Embarrassment-0.070-0.260.000-0.100.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.551+0.600.336+0.360.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.004-0.050.004-0.050.0280.539
19Fearemo_Fear-0.030-0.070.043+0.050.0150.610
20Helplessnessemo_Helplessness0.041+0.060.027+0.04-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.012-0.14-0.006-0.160.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.172+0.030.092-0.080.1490.705
23Infatuationemo_Infatuation-0.018-0.010.010+0.05-0.0150.533
24Interestemo_Interest1.495-0.101.581-0.011.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness-0.035-0.02-0.048-0.04-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.039-0.050.074+0.030.0590.442
27Longingemo_Longing0.030-0.090.072-0.020.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.121+0.120.274+0.410.0550.534
29Painemo_Pain0.021+0.040.028+0.05-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy-0.015-0.050.028+0.020.0170.614
31Prideemo_Pride0.019+0.030.036+0.070.0020.509
32Reliefemo_Relief0.013-0.030.026-0.010.0330.681
33Sadnessemo_Sadness0.011+0.01-0.019-0.020.0000.759
34Sexual Lustemo_Sexual_Lust-0.054-0.170.002-0.060.0300.494
35Shameemo_Shame0.001-0.02-0.006-0.040.0130.510
36Sournessemo_Sourness0.057+0.140.284+0.66-0.0050.439
37Teasingemo_Teasing0.149+0.270.326+0.600.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.081+0.060.042+0.000.0410.697
39Triumphemo_Triumph0.003+0.020.023+0.06-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg3.351+0.303.577+0.482.9841.238
41Arousal (AROU)vn_AROU_reg1.825-0.471.662-0.612.3591.147
42Arousal shift (ARSH)vn_ARSH_reg2.355-0.262.476-0.122.5800.867
43Vocal attack (ATCK)vn_ATCK_reg1.685-0.371.612-0.422.1621.295
44Background noise (BKGN)vn_BKGN_reg2.878+0.232.876+0.232.7030.761
45Brightness (BRGT)vn_BRGT_reg1.150-1.041.026-1.191.9910.813
46Speech chunking (CHNK)vn_CHNK_reg2.213+0.302.060+0.171.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg1.388-0.671.555-0.492.0180.945
48Cognitive load (COGL)vn_COGL_reg2.738+0.612.474+0.362.0941.061
49Dynamic arc (DARC)vn_DARC_reg1.155-0.331.104-0.371.6151.380
50Disfluency (DFLU)vn_DFLU_reg2.940+0.383.027+0.462.4881.184
51Emphasis (EMPH)vn_EMPH_reg2.326-0.182.112-0.332.6071.524
52Esthetics (ESTH)vn_ESTH_reg2.144-0.332.174-0.292.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.226+0.430.299+0.660.0860.322
54Focus (FOCS)vn_FOCS_reg2.042-0.272.174-0.152.3441.102
55Fullness (FULL)vn_FULL_reg1.710-0.081.727-0.061.7780.889
56Perceived gender (GEND)vn_GEND_reg4.615+0.354.951+0.564.0551.588
57Harmonicity (HARM)vn_HARM_reg1.860-0.741.741-0.882.4800.836
58Metallic character (METL)vn_METL_reg2.956+1.392.956+1.391.6210.958
59Pitch range (RANG)vn_RANG_reg2.602-0.352.611-0.342.9140.897
60Recording quality (RCQL)vn_RCQL_reg2.330-0.482.307-0.502.7540.888
61Vocal register (REGS)vn_REGS_reg0.245-0.750.225-0.771.1981.268
62Respiration (RESP)vn_RESP_reg2.278-0.382.086-0.562.6801.056
63Roughness (ROUG)vn_ROUG_reg2.341+0.352.142+0.122.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg3.010+1.043.049+1.082.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg1.888-0.321.801-0.422.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg1.846-0.061.598-0.311.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg2.213+0.932.285+1.011.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg3.209+0.403.196+0.392.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg2.693+0.692.597+0.562.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg3.043+1.213.068+1.242.0700.803
71Smoothness (SMTH)vn_SMTH_reg2.134-0.012.017-0.112.1431.093
72Stance (STNC)vn_STNC_reg2.364-0.242.428-0.182.6331.137
73Speech structure (STRU)vn_STRU_reg2.746+0.642.564+0.452.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg2.163+0.562.236+0.611.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg2.112+0.982.019+0.891.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg2.423+0.622.519+0.681.5601.401
77Casual style (S_CASU)vn_S_CASU_reg3.384-0.323.299-0.393.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg3.880+0.983.941+1.022.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg1.756-0.841.658-0.932.7211.148
80Formal style (S_FORM)vn_S_FORM_reg1.742+0.541.663+0.461.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg3.051+0.182.768-0.042.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg2.748+1.002.299+0.691.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.553+0.260.450+0.070.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg2.969+0.472.944+0.452.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg0.906-0.241.006-0.161.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg3.207+0.533.169+0.502.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg2.070+1.161.880+1.010.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg2.997+0.873.203+0.991.5971.618
89Tempo (TEMP)vn_TEMP_reg1.599-0.441.444-0.572.1171.190
90Tension (TENS)vn_TENS_reg2.137+0.681.846+0.341.5590.852
91Valence (VALN)vn_VALN_reg2.848+0.473.003+0.602.3281.118
92Valence shift (VALS)vn_VALS_reg3.077+0.523.163+0.602.5181.079
93Velocity flux (VFLX)vn_VFLX_reg1.908-0.291.896-0.302.2561.208
94Volatility (VOLT)vn_VOLT_reg2.128-0.522.098-0.552.5860.886
95Vulnerability (VULN)vn_VULN_reg1.753-0.911.695-0.972.6480.984
96Warmth (WARM)vn_WARM_reg2.876+0.332.948+0.412.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_62.414+0.052.557+0.142.3321.621
98Vocal-burst blend (0–10)blend_0_102.291-1.601.888-1.756.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.663-1.280.632-1.371.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.354+0.353.365+0.393.2500.293
101Arousaleiv_extra:Arousal0.988-0.690.999-0.671.2890.435
102Authenticityeiv_extra:Authenticity2.609-0.382.614-0.342.6670.153
103Background Noiseeiv_extra:Background_Noise0.532-1.040.387-1.590.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.227+0.621.198+0.521.0400.303
105Gendereiv_extra:Gender-0.783-0.16-0.798-0.18-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.871+0.171.895+0.271.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive1.481-0.811.739-0.261.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.213+0.212.263+0.362.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous1.155+0.661.258+0.960.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.261-0.060.355+0.230.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.317-0.180.401+0.080.3760.324
112Valenceeiv_extra:Valence-0.014-0.05-0.005-0.030.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached2.108+0.901.975+0.601.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold0.112+0.200.180+0.430.0510.300
115score background qualityeiv_extra:score_background_quality3.599+0.433.694+0.573.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.644+0.454.680+0.574.5070.303
117score overall qualityeiv_extra:score_overall_quality3.020+0.733.052+0.792.6520.507
118score speech qualityeiv_extra:score_speech_quality1.921+0.781.945+0.941.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.097+0.295.490+0.684.7991.018
120Content usefulness (CU)audiobox:CU5.666+0.025.541-0.115.6491.001
121Production complexity (PC)audiobox:PC1.669-0.411.694-0.392.3311.617
122Production quality (PQ)audiobox:PQ7.213+0.197.151+0.127.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw4.134+0.354.155+0.383.9000.669
124Background quality, rawdnsmos:BAK_raw3.806+0.533.999+0.703.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.599+0.633.723+0.793.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.616+0.263.628+0.283.4880.501
127Background quality, calibrateddnsmos:BAK3.731+0.373.898+0.553.3890.928
128Overall quality, calibrateddnsmos:OVRL3.162+0.533.262+0.702.8400.606
129P.808 MOS estimatednsmos:P808_MOS4.086+1.344.055+1.283.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.527+0.810.377+0.580.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.665-0.510.721-0.410.9470.554
132Agevoiceclap_attribute:Age3.306+0.143.093-0.253.2290.547
133Amusementvoiceclap_attribute:Amusement0.530+0.240.361-0.050.3920.572
134Angervoiceclap_attribute:Anger0.686-0.610.688-0.600.9920.506
135Arousalvoiceclap_attribute:Arousal1.324-0.151.305-0.181.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.593+0.180.596+0.180.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.654-0.702.655-0.692.7700.167
138Awevoiceclap_attribute:Awe0.382-0.600.457-0.320.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.462-0.400.510-0.220.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.313-1.190.327-1.130.5920.236
141Concentrationvoiceclap_attribute:Concentration1.192-0.041.080-0.251.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.505+0.931.478+0.871.0690.468
143Confusionvoiceclap_attribute:Confusion0.934+0.870.979+1.000.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.836+0.140.895+0.280.7800.406
145Contemptvoiceclap_attribute:Contempt0.582-0.370.600-0.320.7090.339
146Contentmentvoiceclap_attribute:Contentment0.394-0.630.407-0.590.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.570-0.650.470-1.000.7600.291
148Disgustvoiceclap_attribute:Disgust0.514-0.530.486-0.670.6250.207
149Distressvoiceclap_attribute:Distress0.563-1.040.524-1.150.9270.350
150Doubtvoiceclap_attribute:Doubt0.658+0.660.653+0.630.5350.188
151Elationvoiceclap_attribute:Elation0.471-0.170.465-0.190.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.329+0.070.314+0.020.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.711+0.180.697+0.130.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.471+0.040.429-0.130.4620.261
155Fearvoiceclap_attribute:Fear0.333-1.340.364-1.200.6180.212
156Gendervoiceclap_attribute:Gender-0.637-0.48-0.783-0.63-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.446-0.790.413-0.910.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.960+0.322.009+0.451.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.626-0.200.596-0.270.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability0.869-0.340.902-0.281.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.430-0.540.556-0.160.6080.332
162Interestvoiceclap_attribute:Interest1.452-0.331.450-0.341.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.583+0.170.684+0.390.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.372-0.610.441-0.330.5230.245
165Longingvoiceclap_attribute:Longing0.510-0.130.510-0.130.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.521-0.900.644-0.550.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive1.521-0.431.622-0.261.7750.594
168Painvoiceclap_attribute:Pain0.227-0.710.228-0.700.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.367-0.450.393-0.370.5030.301
170Pridevoiceclap_attribute:Pride0.399-1.090.459-0.920.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.232-0.252.201-0.342.3130.328
172Reliefvoiceclap_attribute:Relief0.736-0.700.658-0.881.0250.415
173Sadnessvoiceclap_attribute:Sadness0.578-0.560.514-0.760.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous1.159+0.631.131+0.580.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.530-0.250.629-0.020.6350.419
176Shamevoiceclap_attribute:Shame0.425-0.360.440-0.310.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.085-0.210.078-0.230.1600.360
178Sournessvoiceclap_attribute:Sourness0.504-0.690.528-0.610.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.438-0.520.469-0.430.6230.359
180Teasingvoiceclap_attribute:Teasing0.687+0.390.630+0.240.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.269-0.221.143-0.451.3840.529
182Triumphvoiceclap_attribute:Triumph0.367-0.970.422-0.820.7160.359
183Valencevoiceclap_attribute:Valence-0.095-0.14-0.324-0.50-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached2.041+0.721.914+0.501.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold-0.004+0.16-0.006+0.15-0.0660.400
186durationvoiceclap_attribute:duration6.761-0.657.454-0.5410.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.427+0.373.430+0.373.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.710+0.374.700+0.304.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.808+0.422.799+0.402.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.975+1.211.952+0.881.8910.069
191talking speedvoiceclap_attribute:talking_speed11.336+0.6111.160+0.579.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
0-0.1098-0.1160-0.0745
10.05440.08480.1359
20.12950.09820.1670
30.07980.02790.1675
4-0.1098-0.0333-0.0669
5-0.1227-0.0848-0.0948
60.01890.0753-0.0062
7-0.0663-0.0688-0.0685
80.04070.08350.0167
9-0.04690.01440.1032
100.07350.07400.1071
11-0.0047-0.03890.0715
12-0.02760.05930.0307
13-0.03210.0041-0.0598
140.16470.16790.1343
15-0.1155-0.1052-0.0671
16-0.1227-0.0827-0.0927
17-0.1129-0.1333-0.0779
18-0.0570-0.1103-0.0569
19-0.0875-0.0710-0.0708
20-0.1025-0.0883-0.0750
21-0.0678-0.0844-0.0688
220.08130.12380.0551
23-0.0089-0.0366-0.0876
24-0.0466-0.0632-0.0605
25-0.0399-0.0110-0.0490
26-0.0766-0.03810.0064
27-0.0813-0.0701-0.0698
280.06370.0255-0.0223
29-0.0510-0.0413-0.0802
30-0.1036-0.1108-0.0806
310.03990.07440.0331
32-0.0694-0.0744-0.0721
33-0.0155-0.0387-0.0818
34-0.0533-0.0917-0.0671
35-0.0953-0.1060-0.0865
360.00720.0032-0.0318
37-0.1088-0.1480-0.0887
380.08860.01380.0617
39-0.0787-0.0891-0.0756
40-0.0741-0.0602-0.0722
410.0183-0.03760.1426
42-0.1155-0.0844-0.0756
43-0.0202-0.0359-0.0053
44-0.0787-0.0610-0.0719
450.10510.08650.0811
460.00300.0167-0.0706
47-0.0249-0.05800.0559
480.02250.06710.0462
490.29420.27350.2230
50-0.0730-0.0714-0.0766
510.20920.18090.1884
520.0216-0.0718-0.0806
53-0.0311-0.0684-0.0146
540.0518-0.00820.0018
550.05960.04390.0963
560.03550.13410.1826
57-0.01020.03940.0892
580.07870.14970.1351
590.02460.1073-0.0277
60-0.0963-0.0740-0.0222
610.07920.05410.0217
620.0093-0.00830.0019
63-0.0378-0.04220.0028
640.0471-0.00360.0291
650.18020.18870.2296
660.04560.11420.0563
67-0.0337-0.0179-0.0215
680.06030.11340.1180
690.27550.23890.2249
70-0.0891-0.1177-0.0776
710.03780.0513-0.0020
720.13000.12640.1641
73-0.03550.01560.0642
740.04170.03510.0648
75-0.0153-0.0675-0.0909
760.09680.10730.1110
77-0.0870-0.0645-0.0684
78-0.1098-0.0788-0.0667
790.11030.0320-0.0112
800.09890.0861-0.0185
810.19470.09090.0983
82-0.0299-0.0457-0.0030
830.04350.13670.1636
84-0.0417-0.0476-0.0502
850.0264-0.0011-0.0445
860.01350.09130.0422
87-0.0860-0.0457-0.0199
880.08290.11340.0771
890.00460.0567-0.0018
900.08860.04330.1223
91-0.04400.0061-0.0688
920.23410.29250.2214
93-0.0673-0.0662-0.0536
94-0.0300-0.0762-0.0451
950.11960.09000.1032
96-0.0001-0.03010.0073
970.15740.10260.1226
98-0.0777-0.0576-0.0757
99-0.02890.01040.0060
1000.17710.17830.1633
1010.06530.0376-0.0724
102-0.0500-0.0357-0.0381
1030.03990.02960.0377
104-0.0629-0.1125-0.0732
1050.01240.0387-0.0677
1060.09840.05430.1074
1070.01320.04610.0709
108-0.1005-0.1047-0.0812
109-0.02460.02700.1039
110-0.0073-0.0309-0.0693
1110.04920.05630.0266
112-0.0461-0.1073-0.0154
113-0.0603-0.0602-0.0568
114-0.0179-0.0491-0.0806
1150.0337-0.0163-0.0388
1160.21030.15410.1216
117-0.1150-0.1095-0.0872
118-0.02320.06230.1112
119-0.0432-0.0666-0.0724
120-0.1155-0.0448-0.0770
1210.02810.01340.0099
1220.01370.01640.0241
123-0.0256-0.03440.0280
124-0.0077-0.01220.0978
125-0.1051-0.0961-0.0746
1260.03470.00650.0852
1270.02620.01040.0323
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.02030.0190-0.0201
1-0.0599-0.0611-0.0442
2-0.0388-0.0233-0.0561
3-0.0081-0.0402-0.0865
4-0.0228-0.0543-0.0415
50.02540.08220.0511
6-0.1080-0.1282-0.0669
7-0.0382-0.0490-0.0462
8-0.0130-0.00560.0691
90.02620.03810.0200
10-0.0312-0.0317-0.0399
11-0.0373-0.0539-0.0617
120.02590.04320.0807
130.02650.03580.0608
14-0.0852-0.0890-0.0651
150.10500.11090.1246
160.0350-0.03220.0815
17-0.02950.0040-0.0548
180.14680.17280.2107
19-0.0405-0.0728-0.0724
20-0.0348-0.0551-0.0457
210.0126-0.0090-0.0192
22-0.0133-0.0127-0.0324
23-0.0407-0.0330-0.0441
24-0.0517-0.0709-0.0389
250.01140.00770.0533
26-0.0696-0.0668-0.0503
27-0.0536-0.0528-0.0371
28-0.0256-0.0245-0.0494
29-0.0667-0.0585-0.0439
30-0.0399-0.0272-0.0397
310.11730.13880.0958
32-0.0603-0.0163-0.0469
33-0.0418-0.0255-0.0410
34-0.0721-0.0943-0.0545
350.07760.04940.0078
360.04110.00120.0365
370.06620.02720.0255
380.04200.08750.0844
390.02480.0379-0.0499
40-0.0312-0.0207-0.0213
41-0.0416-0.0255-0.0524
42-0.0688-0.0487-0.0453
430.06370.01280.0480
44-0.0700-0.0102-0.0549
450.0304-0.0290-0.0139
46-0.0557-0.0185-0.0660
47-0.1050-0.0732-0.0597
480.06200.05430.1153
49-0.0250-0.0025-0.0057
500.08350.07690.0590
51-0.0574-0.0709-0.0462
52-0.0062-0.0377-0.0408
53-0.0437-0.0279-0.0441
54-0.0462-0.0192-0.0771
550.02250.0195-0.0142
560.01720.02180.0155
57-0.1198-0.1215-0.0533
580.13080.14710.1318
590.0157-0.03300.0039
60-0.0243-0.0082-0.0478
610.03920.08860.0643
62-0.0390-0.0268-0.0401
63-0.0540-0.0532-0.0496
64-0.0373-0.0468-0.0508
650.0189-0.0174-0.0557
66-0.0233-0.0253-0.0575
670.09910.03130.0163
68-0.01310.04260.0500
690.03840.00580.0481
70-0.0141-0.0099-0.0662
71-0.0751-0.0407-0.0478
72-0.0962-0.0924-0.0505
73-0.0439-0.0634-0.0814
74-0.0742-0.0747-0.0492
750.08560.09010.0890
76-0.0239-0.0581-0.0434
770.06790.02600.0274
780.00560.08070.0158
79-0.0502-0.0521-0.0488
800.07170.05580.1559
810.02270.0028-0.0546
820.20250.14110.1150
83-0.0553-0.0490-0.0364
84-0.0137-0.0509-0.0577
85-0.0641-0.0785-0.0487
86-0.0755-0.0849-0.0435
870.0085-0.0573-0.0389
880.08100.06710.0095
89-0.0772-0.0447-0.0440
90-0.0527-0.0438-0.0483
910.00020.0558-0.0573
92-0.0540-0.0660-0.0421
93-0.1198-0.0494-0.0553
94-0.0612-0.03750.0043
950.13580.14940.1874
960.06830.08750.2048
970.03060.07090.1140
98-0.1105-0.0890-0.0515
990.08940.02730.0907
100-0.0721-0.0498-0.0549
101-0.0738-0.0656-0.0460
102-0.00070.0034-0.0063
103-0.0789-0.0751-0.0805
104-0.0130-0.02290.0163
105-0.03480.0426-0.0021
106-0.0553-0.0656-0.0375
107-0.0667-0.0875-0.0436
108-0.0662-0.0600-0.0362
109-0.0759-0.0679-0.0468
110-0.0137-0.01650.0224
111-0.0529-0.0524-0.0401
112-0.0230-0.0311-0.0105
113-0.0924-0.0788-0.0830
1140.00030.0200-0.0036
1150.13500.10330.0958
116-0.1055-0.1071-0.0875
1170.04430.00750.0273
118-0.0018-0.0054-0.0558
1190.0163-0.02680.0023
120-0.0316-0.0372-0.0621
121-0.0081-0.0083-0.0524
1220.0091-0.00430.0005
1230.08100.04090.0206
124-0.0772-0.0913-0.0486
125-0.0586-0.0713-0.0473
126-0.0124-0.0604-0.0942
127-0.1080-0.0758-0.0501
128-0.0470-0.0351-0.0257
1290.05480.02470.0339
1300.06960.05320.0690
131-0.0248-0.0421-0.0426
132-0.1029-0.0815-0.0516
133-0.0675-0.0943-0.0482
134-0.0367-0.0468-0.0233
1350.07130.07170.0198
136-0.0534-0.05620.0322
137-0.0561-0.0273-0.0452
138-0.0574-0.0686-0.0429
139-0.0071-0.0004-0.0498
1400.06160.07130.0743
141-0.0941-0.0837-0.0482
142-0.01940.0187-0.0421
143-0.0352-0.00580.1223
1440.05440.08030.0639
1450.00950.0383-0.0316
146-0.0281-0.0262-0.0584
147-0.0201-0.0157-0.0433
148-0.0688-0.0698-0.0537
1490.11390.16290.0546
1500.0688-0.0164-0.0479
151-0.00020.0137-0.0555
152-0.0210-0.0324-0.0540
1530.06070.0702-0.0351
154-0.0206-0.0322-0.0475
155-0.0228-0.0513-0.0635
156-0.0772-0.0822-0.0488
1570.04600.08490.0517
1580.0231-0.0141-0.0606
1590.05020.0055-0.0570
160-0.0188-0.0273-0.0463
161-0.1434-0.0924-0.0496
1620.03190.01900.0702
163-0.00300.0209-0.0287
164-0.0831-0.0637-0.0428
165-0.03690.0383-0.0101
1660.07050.09810.0915
167-0.0409-0.06040.0146
1680.09530.0481-0.0250
1690.06160.05430.0201
170-0.0768-0.0585-0.0456
1710.01200.0237-0.0428
172-0.0094-0.0675-0.0564
173-0.0388-0.1056-0.0571
1740.0213-0.01240.1625
175-0.0840-0.1139-0.0807
176-0.1000-0.0777-0.0414
1770.01500.06040.0142
178-0.00160.0251-0.0432
1790.01630.0686-0.0012
180-0.0227-0.0270-0.0557
1810.12230.11920.0441
1820.01030.0377-0.0421
183-0.0696-0.0935-0.0982
184-0.0254-0.0126-0.0252
1850.01460.00190.0846
186-0.0915-0.0453-0.0587
187-0.00930.01040.0270
188-0.0523-0.0736-0.0457
1890.05060.03320.0210
1900.05290.0321-0.0578
1910.04450.0458-0.0651
1920.07510.04710.1243
1930.12320.17580.1449
194-0.0498-0.02070.0564
1950.12990.1735-0.0463
196-0.0373-0.0626-0.0473
197-0.0219-0.0218-0.0681
198-0.0595-0.0702-0.0471
1990.05400.01980.0751
2000.08400.10110.1353
201-0.0734-0.0637-0.0459
202-0.0738-0.1003-0.0095
203-0.01520.01470.0453
2040.02260.03240.0241
2050.05080.02400.0229
2060.01390.0040-0.0502
2070.0090-0.0322-0.0658
208-0.0118-0.0129-0.0505
209-0.0468-0.0762-0.0712
210-0.0176-0.0178-0.0634
2110.0014-0.00410.0071
212-0.0772-0.0328-0.0712
2130.05340.0081-0.0638
2140.00260.0536-0.0614
215-0.0705-0.0966-0.0409
216-0.01980.0061-0.0433
2170.0123-0.01160.0178
2180.13840.16900.1740
219-0.0570-0.0332-0.0457
2200.0071-0.0521-0.0537
221-0.0451-0.0077-0.0029
222-0.1358-0.0466-0.0571
2230.0087-0.0481-0.0391
2240.17210.14030.1780
225-0.0536-0.0736-0.0405
2260.07800.09540.0946
227-0.00120.0205-0.0626
228-0.0295-0.03040.0216
2290.13840.07390.1037
230-0.0953-0.0818-0.0496
231-0.1072-0.1026-0.0674
232-0.0416-0.0539-0.0668
2330.04600.03410.0133
234-0.0456-0.0417-0.0413
2350.05700.07020.0278
236-0.0018-0.0302-0.0249
237-0.0180-0.0021-0.0497
238-0.0620-0.0404-0.0271
239-0.0831-0.0762-0.0503
240-0.0399-0.0041-0.0497
2410.02780.07620.0786
2420.00920.04430.0429
2430.0193-0.0243-0.0450
244-0.0464-0.0349-0.0518
2450.11900.11470.0270
246-0.0310-0.0532-0.0394
2470.10590.07920.1048
248-0.0449-0.0551-0.0534
2490.08060.08710.0812
Source: Emolia / Emilia source archive · SHA-256 d7247351f449dd22…
e06FrenchEmolia / Emilia

Emolia FR · FR_B00000_S00010_W000021

Natural speech · Original speech · 9.0 s

Alors telles que nous avons mentionnées au début de cette session d'étude, avant que ce gouvernement ne soit en place.

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Concentration
z +1.67 / +1.43raw 1.87 / 1.70
Contemplation
z +0.83 / +1.04raw 0.58 / 0.71
Interest
z +0.61 / +0.43raw 2.14 / 1.98
Pride
z +0.58 / +0.47raw 0.30 / 0.24
Contentment
z +0.54 / +0.25raw 0.28 / 0.13
Anger
z +0.27 / +0.26raw 0.17 / 0.16
Emotional Numbness
z +0.15 / +0.14raw 0.14 / 0.14

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Teacher or didactic style (S_TECH)
z +2.01 / +2.36raw 3.12 / 3.54
Authoritative style (S_AUTH)
z +1.65 / +1.78raw 2.81 / 2.95
Mixed resonance (R_MIXD)
z +1.29 / +1.70raw 2.52 / 2.87
Throat resonance (R_THRT)
z +1.62 / +1.64raw 3.37 / 3.39
Chest resonance (R_CHST)
z +1.28 / +1.57raw 3.22 / 3.46
Perceived voice age (AGEV)
z +1.24 / +1.50raw 4.52 / 4.84
Formal style (S_FORM)
z +1.31 / +1.49raw 2.54 / 2.72

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z +0.92 / +0.70raw 3.82 / 3.46
Vocal-burst blend (0–10)z -1.51 / -1.83raw 2.55 / 1.65
Content enjoyment (CE)z +0.53 / +0.66raw 5.34 / 5.47
Production quality (PQ)z -0.22 / -0.06raw 6.85 / 6.99
Speech signal quality, calibratedz +0.18 / +0.22raw 3.58 / 3.60
Background quality, calibratedz +0.66 / +0.39raw 4.00 / 3.75
Overall quality, calibratedz +0.70 / +0.54raw 3.26 / 3.17
P.808 MOS estimatez +0.89 / +0.96raw 3.86 / 3.90
score content enjoymentz +0.17 / +0.42raw 4.56 / 4.63
score overall qualityz +0.86 / +0.63raw 3.09 / 2.97
Background Noisez -0.33 / -1.08raw 0.72 / 0.52
Tempo (TEMP)z +0.01 / -0.25raw 2.13 / 1.82
Arousal (AROU)z -0.24 / -0.32raw 2.08 / 1.99
Valence (VALN)z +0.64 / +0.71raw 3.04 / 3.12
Recording quality (RCQL)z -0.24 / -0.14raw 2.54 / 2.63
Background noise (BKGN)z -0.53 / -0.37raw 2.30 / 2.42
talking speedz +1.09 / +1.26raw 13.13 / 13.76
Characters per second (CPS): 13.16 raw · z +1.10

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.7980.867
Identity · 250D0.6710.762

Vocal-burst regions

Tiny
Base
0 s9.0 s

Verified burst intervals are unavailable for this clip.

Tiny: 0 regions, count-head estimate 0.10; Base: 0 regions, count-head estimate 0.00.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection-0.013-0.04-0.022-0.050.0070.570
1Amusementemo_Amusement-0.020-0.020.024+0.03-0.0010.885
2Angeremo_Anger0.170+0.270.165+0.260.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.030+0.040.012+0.010.0050.572
4Aweemo_Awe0.008+0.020.027+0.060.0000.427
5Bitternessemo_Bitterness0.023+0.02-0.001-0.040.0130.415
6Concentrationemo_Concentration1.869+1.671.698+1.430.6340.741
7Confusionemo_Confusion-0.043-0.080.005+0.010.0010.540
8Contemplationemo_Contemplation0.577+0.830.706+1.040.0820.598
9Contemptemo_Contempt0.020+0.040.016+0.03-0.0000.464
10Contentmentemo_Contentment0.275+0.540.128+0.250.0000.514
11Disappointmentemo_Disappointment-0.077-0.11-0.061-0.090.0000.700
12Disgustemo_Disgust0.057+0.10-0.004-0.020.0070.489
13Distressemo_Distress0.006+0.01-0.030-0.040.0000.821
14Doubtemo_Doubt-0.034-0.080.008+0.000.0060.466
15Elationemo_Elation0.000-0.020.006-0.010.0120.728
16Embarrassmentemo_Embarrassment-0.056-0.23-0.001-0.100.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.141+0.150.137+0.140.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.008-0.040.051+0.040.0280.539
19Fearemo_Fear0.004-0.02-0.013-0.040.0150.610
20Helplessnessemo_Helplessness-0.053-0.07-0.064-0.09-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.185+0.09-0.044-0.210.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability-0.082-0.33-0.012-0.230.1490.705
23Infatuationemo_Infatuation-0.046-0.06-0.021-0.01-0.0150.533
24Interestemo_Interest2.138+0.611.978+0.431.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness-0.025-0.00-0.021+0.01-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.027-0.070.041-0.040.0590.442
27Longingemo_Longing-0.009-0.150.000-0.140.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.085+0.060.029-0.050.0550.534
29Painemo_Pain0.013+0.030.039+0.07-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy-0.073-0.150.006-0.020.0170.614
31Prideemo_Pride0.297+0.580.241+0.470.0020.509
32Reliefemo_Relief0.024-0.010.054+0.030.0330.681
33Sadnessemo_Sadness-0.046-0.06-0.037-0.050.0000.759
34Sexual Lustemo_Sexual_Lust-0.025-0.11-0.025-0.110.0300.494
35Shameemo_Shame0.010-0.010.015+0.000.0130.510
36Sournessemo_Sourness-0.042-0.08-0.003+0.01-0.0050.439
37Teasingemo_Teasing0.003+0.010.028+0.050.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.038-0.000.083+0.060.0410.697
39Triumphemo_Triumph0.054+0.120.010+0.03-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg4.522+1.244.837+1.502.9841.238
41Arousal (AROU)vn_AROU_reg2.080-0.241.988-0.322.3591.147
42Arousal shift (ARSH)vn_ARSH_reg2.614+0.042.591+0.012.5800.867
43Vocal attack (ATCK)vn_ATCK_reg2.535+0.292.345+0.142.1621.295
44Background noise (BKGN)vn_BKGN_reg2.303-0.532.420-0.372.7030.761
45Brightness (BRGT)vn_BRGT_reg2.432+0.542.606+0.761.9910.813
46Speech chunking (CHNK)vn_CHNK_reg2.176+0.272.096+0.201.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg2.208+0.202.273+0.272.0180.945
48Cognitive load (COGL)vn_COGL_reg3.113+0.963.158+1.002.0941.061
49Dynamic arc (DARC)vn_DARC_reg1.802+0.141.831+0.161.6151.380
50Disfluency (DFLU)vn_DFLU_reg3.618+0.953.419+0.792.4881.184
51Emphasis (EMPH)vn_EMPH_reg3.368+0.503.356+0.492.6071.524
52Esthetics (ESTH)vn_ESTH_reg2.541+0.182.580+0.232.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg-0.030-0.360.002-0.260.0860.322
54Focus (FOCS)vn_FOCS_reg2.983+0.582.742+0.362.3441.102
55Fullness (FULL)vn_FULL_reg2.386+0.682.648+0.981.7780.889
56Perceived gender (GEND)vn_GEND_reg4.696+0.404.939+0.564.0551.588
57Harmonicity (HARM)vn_HARM_reg2.145-0.402.174-0.372.4800.836
58Metallic character (METL)vn_METL_reg2.963+1.403.023+1.461.6210.958
59Pitch range (RANG)vn_RANG_reg2.407-0.562.484-0.482.9140.897
60Recording quality (RCQL)vn_RCQL_reg2.544-0.242.629-0.142.7540.888
61Vocal register (REGS)vn_REGS_reg0.564-0.500.412-0.621.1981.268
62Respiration (RESP)vn_RESP_reg2.439-0.232.690+0.012.6801.056
63Roughness (ROUG)vn_ROUG_reg2.636+0.682.564+0.602.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg3.215+1.283.461+1.572.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg2.248+0.122.182+0.042.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg2.510+0.602.387+0.481.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg2.521+1.292.867+1.701.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg4.302+1.444.285+1.432.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg3.236+1.423.172+1.342.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg3.370+1.623.389+1.642.0700.803
71Smoothness (SMTH)vn_SMTH_reg2.046-0.092.128-0.012.1431.093
72Stance (STNC)vn_STNC_reg3.237+0.533.059+0.372.6331.137
73Speech structure (STRU)vn_STRU_reg2.434+0.312.356+0.232.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg0.564-0.570.758-0.431.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg2.810+1.652.948+1.781.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg1.857+0.212.099+0.391.5601.401
77Casual style (S_CASU)vn_S_CASU_reg3.364-0.343.132-0.523.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg3.025+0.332.852+0.202.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg1.974-0.652.134-0.512.7211.148
80Formal style (S_FORM)vn_S_FORM_reg2.537+1.312.719+1.491.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg3.456+0.493.625+0.622.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg1.752+0.311.755+0.311.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.691+0.500.763+0.620.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg2.842+0.372.903+0.422.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg0.562-0.500.432-0.611.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg2.528+0.032.569+0.062.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg3.117+2.013.535+2.360.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg1.790+0.121.924+0.201.5971.618
89Tempo (TEMP)vn_TEMP_reg2.130+0.011.821-0.252.1171.190
90Tension (TENS)vn_TENS_reg1.994+0.512.092+0.631.5590.852
91Valence (VALN)vn_VALN_reg3.042+0.643.122+0.712.3281.118
92Valence shift (VALS)vn_VALS_reg3.759+1.153.862+1.252.5181.079
93Velocity flux (VFLX)vn_VFLX_reg2.256-0.002.489+0.192.2561.208
94Volatility (VOLT)vn_VOLT_reg2.301-0.322.186-0.452.5860.886
95Vulnerability (VULN)vn_VULN_reg1.978-0.681.819-0.842.6480.984
96Warmth (WARM)vn_WARM_reg2.740+0.182.783+0.232.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_63.823+0.923.459+0.702.3321.621
98Vocal-burst blend (0–10)blend_0_102.546-1.511.654-1.836.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.910-0.550.756-1.001.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.426+0.603.424+0.593.2500.293
101Arousaleiv_extra:Arousal1.125-0.381.022-0.611.2890.435
102Authenticityeiv_extra:Authenticity2.660-0.052.665-0.022.6670.153
103Background Noiseeiv_extra:Background_Noise0.721-0.330.521-1.080.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant0.909-0.430.898-0.471.0400.303
105Gendereiv_extra:Gender-1.137-0.52-1.106-0.49-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched2.192+1.462.138+1.241.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive1.637-0.471.571-0.611.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.288+0.432.359+0.652.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous0.860-0.200.820-0.320.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.103-0.560.065-0.670.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.456+0.250.429+0.170.3760.324
112Valenceeiv_extra:Valence0.063+0.150.047+0.110.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached2.375+1.512.375+1.511.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold0.126+0.250.027-0.080.0510.300
115score background qualityeiv_extra:score_background_quality3.738+0.633.654+0.513.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.559+0.174.634+0.424.5070.303
117score overall qualityeiv_extra:score_overall_quality3.088+0.862.971+0.632.6520.507
118score speech qualityeiv_extra:score_speech_quality1.937+0.891.946+0.951.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.336+0.535.475+0.664.7991.018
120Content usefulness (CU)audiobox:CU5.484-0.165.573-0.085.6491.001
121Production complexity (PC)audiobox:PC1.890-0.271.738-0.372.3311.617
122Production quality (PQ)audiobox:PQ6.854-0.226.988-0.067.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw4.066+0.254.105+0.313.9000.669
124Background quality, rawdnsmos:BAK_raw4.100+0.793.837+0.563.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.719+0.783.599+0.633.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.578+0.183.597+0.223.4880.501
127Background quality, calibrateddnsmos:BAK4.003+0.663.749+0.393.3890.928
128Overall quality, calibrateddnsmos:OVRL3.264+0.703.169+0.542.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.861+0.893.899+0.963.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.091+0.140.002+0.000.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.877-0.130.652-0.530.9470.554
132Agevoiceclap_attribute:Age3.532+0.553.398+0.313.2290.547
133Amusementvoiceclap_attribute:Amusement0.303-0.160.182-0.370.3920.572
134Angervoiceclap_attribute:Anger0.888-0.210.801-0.380.9920.506
135Arousalvoiceclap_attribute:Arousal1.556+0.221.415-0.011.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.577+0.140.303-0.430.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.829+0.352.824+0.322.7700.167
138Awevoiceclap_attribute:Awe0.743+0.790.438-0.390.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.466-0.380.383-0.700.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.539-0.230.526-0.280.5920.236
141Concentrationvoiceclap_attribute:Concentration1.639+0.801.716+0.941.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.362+0.621.287+0.471.0690.468
143Confusionvoiceclap_attribute:Confusion0.606-0.110.425-0.650.6430.334
144Contemplationvoiceclap_attribute:Contemplation1.111+0.821.122+0.840.7800.406
145Contemptvoiceclap_attribute:Contempt0.635-0.220.564-0.430.7090.339
146Contentmentvoiceclap_attribute:Contentment0.483-0.340.462-0.410.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.766+0.020.624-0.470.7600.291
148Disgustvoiceclap_attribute:Disgust0.604-0.100.498-0.610.6250.207
149Distressvoiceclap_attribute:Distress0.909-0.050.620-0.880.9270.350
150Doubtvoiceclap_attribute:Doubt0.412-0.650.325-1.120.5350.188
151Elationvoiceclap_attribute:Elation0.585+0.120.489-0.130.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.273-0.110.150-0.500.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.584-0.300.694+0.110.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.308-0.590.274-0.720.4620.261
155Fearvoiceclap_attribute:Fear0.335-1.330.239-1.790.6180.212
156Gendervoiceclap_attribute:Gender-1.077-0.92-1.281-1.12-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.559-0.370.384-1.020.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched2.292+1.212.275+1.161.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.840+0.340.818+0.280.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability0.996-0.100.817-0.431.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.269-1.020.149-1.380.6080.332
162Interestvoiceclap_attribute:Interest1.899+0.681.911+0.711.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.204-0.640.202-0.650.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.321-0.820.305-0.890.5230.245
165Longingvoiceclap_attribute:Longing0.460-0.360.329-0.950.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.599-0.680.604-0.660.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive1.938+0.271.758-0.031.7750.594
168Painvoiceclap_attribute:Pain0.200-0.830.082-1.370.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.250-0.840.196-1.020.5030.301
170Pridevoiceclap_attribute:Pride0.907+0.360.818+0.110.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.332+0.062.400+0.262.3130.328
172Reliefvoiceclap_attribute:Relief0.641-0.930.442-1.401.0250.415
173Sadnessvoiceclap_attribute:Sadness0.716-0.110.498-0.810.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.802-0.000.787-0.030.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.154-1.150.082-1.320.6350.419
176Shamevoiceclap_attribute:Shame0.644+0.420.480-0.170.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.080-0.220.024-0.380.1600.360
178Sournessvoiceclap_attribute:Sourness0.561-0.500.419-0.990.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.685+0.170.666+0.120.6230.359
180Teasingvoiceclap_attribute:Teasing0.184-0.940.134-1.070.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.376-0.011.052-0.631.3840.529
182Triumphvoiceclap_attribute:Triumph0.665-0.140.546-0.470.7160.359
183Valencevoiceclap_attribute:Valence-0.408-0.63-0.352-0.54-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached2.230+1.062.356+1.291.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold0.103+0.420.101+0.42-0.0660.400
186durationvoiceclap_attribute:duration9.161-0.288.840-0.3310.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.579+0.673.585+0.683.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.641-0.104.679+0.164.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.852+0.522.845+0.512.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.962+1.021.975+1.211.8910.069
191talking speedvoiceclap_attribute:talking_speed13.131+1.0913.762+1.269.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
0-0.0984-0.1097-0.1041
10.34760.21400.0931
2-0.0451-0.0472-0.0577
3-0.0773-0.0594-0.0694
40.06540.0392-0.0316
5-0.1375-0.1433-0.1324
6-0.0692-0.0635-0.0948
7-0.0554-0.0440-0.0608
8-0.1432-0.1061-0.0940
9-0.0802-0.0757-0.0365
10-0.0702-0.0825-0.0562
11-0.0898-0.1233-0.0979
12-0.0004-0.0635-0.0539
13-0.0874-0.0943-0.0985
14-0.0101-0.00630.0070
150.03680.0131-0.0224
160.02590.06210.1078
17-0.0735-0.0469-0.0507
180.08020.05460.1425
19-0.0449-0.0278-0.0075
20-0.0516-0.0825-0.1048
21-0.0754-0.0925-0.0963
22-0.1022-0.1011-0.0984
23-0.1490-0.1487-0.1238
24-0.0275-0.02060.0004
250.12700.13510.0563
260.09020.12060.1494
27-0.0793-0.0703-0.0545
28-0.0812-0.0698-0.0699
290.03750.05370.0126
300.02200.05580.0207
310.01060.05010.1107
32-0.0879-0.1007-0.1008
33-0.0461-0.0255-0.0070
34-0.0310-0.0789-0.0798
35-0.0344-0.0406-0.1020
360.06400.05150.0582
37-0.0580-0.0676-0.1043
38-0.0103-0.0286-0.0017
39-0.0855-0.0948-0.1056
40-0.1046-0.0934-0.1010
41-0.0344-0.0424-0.0224
420.03750.09520.1188
430.07590.04130.0654
440.06020.03290.0725
45-0.0520-0.01130.0308
460.17380.24300.1063
47-0.0998-0.0839-0.1058
480.12510.10200.0571
490.03100.0114-0.0171
50-0.1337-0.1061-0.1071
510.07500.0438-0.0242
52-0.0668-0.0875-0.0658
530.0149-0.0229-0.0406
540.02290.05980.1622
550.0277-0.02380.0768
560.0133-0.0447-0.0179
57-0.0139-0.0458-0.0740
58-0.0118-0.0125-0.0438
59-0.0542-0.0320-0.0123
60-0.1151-0.1161-0.1010
61-0.0224-0.0193-0.0964
62-0.1027-0.1034-0.0979
630.00440.02490.1060
64-0.1155-0.1025-0.1062
650.05730.07800.0953
660.18810.21220.2052
67-0.0797-0.0617-0.0972
68-0.00680.03900.0152
69-0.0523-0.0753-0.0959
70-0.0246-0.0639-0.0922
71-0.0439-0.0703-0.1132
72-0.1289-0.1206-0.1013
73-0.1003-0.1056-0.0780
740.12320.12700.1049
75-0.0473-0.00480.0393
760.19670.18680.1857
77-0.0084-0.00170.0528
780.00250.0278-0.0321
79-0.0399-0.0562-0.0908
80-0.01670.03330.0723
81-0.0006-0.0349-0.0219
82-0.0950-0.1120-0.1002
830.10930.07890.0259
84-0.0437-0.0444-0.0138
85-0.0640-0.0929-0.0538
86-0.0286-0.0916-0.0673
870.09170.13060.2100
880.16710.19130.1740
890.00690.0499-0.0193
90-0.0463-0.0553-0.1135
910.09210.12970.0790
920.07730.0408-0.0177
93-0.0134-0.03130.0464
94-0.1065-0.0934-0.1066
950.13080.05280.0629
96-0.0423-0.02850.0395
970.01420.01130.0128
980.01130.03150.1442
99-0.0167-0.0347-0.1003
1000.19000.24660.2001
1010.03890.02700.0112
102-0.0955-0.0744-0.1120
103-0.0195-0.0311-0.0804
104-0.0726-0.0807-0.1024
1050.0212-0.0247-0.0300
1060.0282-0.01670.1048
1070.14510.16230.1656
108-0.0387-0.05280.0266
109-0.0042-0.0212-0.0459
110-0.00110.00620.0144
111-0.0220-0.01620.0168
1120.06020.03740.0089
1130.08740.04490.1199
114-0.0807-0.0866-0.1127
1150.0377-0.0126-0.0039
1160.25210.23580.1210
117-0.1356-0.1029-0.0733
1180.12410.08710.0472
1190.11890.10290.0800
120-0.0592-0.0338-0.1077
1210.08260.02550.0493
122-0.0280-0.02630.0345
1230.15280.19410.1426
124-0.0060-0.0576-0.0468
1250.04990.06390.0453
126-0.0759-0.0852-0.0902
1270.09600.07210.0673
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.00280.0341-0.0459
10.01140.07150.1121
20.0431-0.0197-0.0560
3-0.0262-0.0332-0.0920
40.04930.07480.0782
50.05810.03270.0475
60.0035-0.01200.0559
70.0150-0.01540.0031
8-0.0968-0.0454-0.0540
9-0.0293-0.0767-0.0517
10-0.0306-0.0880-0.0508
11-0.0845-0.0457-0.0656
12-0.00190.04190.1200
130.04440.03080.0313
140.05590.05720.0180
150.04050.09040.0447
16-0.0607-0.0913-0.0700
170.00560.02240.0535
18-0.0086-0.02980.0204
190.0332-0.0003-0.0435
20-0.0253-0.0814-0.0485
210.02820.0235-0.0238
22-0.00450.0026-0.0503
23-0.0963-0.0791-0.0469
24-0.01470.02990.0624
25-0.0042-0.01970.0214
260.0126-0.0379-0.0534
27-0.00200.01920.0645
28-0.0019-0.0175-0.0051
290.09630.12330.0289
300.27620.23250.1484
310.15750.12520.2035
320.04510.06020.1194
33-0.0457-0.0147-0.0020
34-0.0400-0.04640.0148
350.06730.07150.0460
360.01970.0304-0.0423
370.17860.19020.1743
380.11170.08990.1122
39-0.0195-0.03770.0185
40-0.0131-0.02940.0444
410.06910.03740.0235
42-0.0545-0.0819-0.0482
430.16980.10260.1307
440.04420.02440.0412
45-0.0133-0.0374-0.0288
46-0.00680.0417-0.0164
47-0.1122-0.1214-0.0635
480.0365-0.01030.0356
49-0.01660.05110.0511
50-0.0267-0.0189-0.0074
51-0.02320.02250.0667
52-0.03210.02120.0687
530.02740.04120.0470
540.11960.14970.1288
55-0.0374-0.0791-0.0536
56-0.0479-0.0148-0.0569
570.02350.09320.0354
580.08360.0346-0.0071
590.05170.07810.1157
60-0.0321-0.0522-0.0271
61-0.0012-0.0617-0.0450
620.02160.0265-0.0067
63-0.0664-0.0993-0.0526
64-0.0011-0.0195-0.0540
65-0.1496-0.1261-0.0591
66-0.0682-0.0506-0.0610
670.07870.01750.0200
68-0.0119-0.0101-0.0446
690.02740.04640.0128
70-0.0288-0.02890.0030
710.03560.0182-0.0508
72-0.00130.00980.0482
730.0241-0.0228-0.0039
740.0046-0.02690.0267
75-0.0477-0.00230.0793
760.08450.00730.0345
770.00200.02890.0524
780.03650.0617-0.0172
79-0.0189-0.0221-0.0518
80-0.04020.05700.0439
81-0.0471-0.0426-0.0580
82-0.0139-0.0421-0.0465
830.02200.04050.0513
84-0.0853-0.0974-0.0729
85-0.0955-0.0626-0.0518
860.0112-0.02340.0548
870.05230.01100.1209
880.0096-0.0106-0.0038
890.02280.0286-0.0390
90-0.0968-0.0555-0.0513
91-0.0552-0.04330.0041
920.0020-0.0532-0.0447
930.15400.14590.1815
94-0.02930.0147-0.0171
95-0.0686-0.0081-0.0540
96-0.0352-0.01970.0033
97-0.02640.02350.0097
980.01980.03320.0345
99-0.02250.02850.0511
100-0.0796-0.1487-0.0584
101-0.0078-0.02450.0281
102-0.0501-0.0466-0.0578
103-0.03260.02440.0590
104-0.0490-0.0520-0.0567
1050.01390.0687-0.0155
1060.11960.03250.0184
107-0.0862-0.0490-0.0464
1080.0161-0.0207-0.0059
109-0.00540.0148-0.0411
1100.00550.0019-0.0207
1110.01050.0419-0.0136
1120.0378-0.0212-0.0418
113-0.0444-0.0384-0.0882
114-0.0809-0.0513-0.0576
1150.14520.16190.1388
116-0.0787-0.0701-0.0929
1170.09330.05220.0403
118-0.0493-0.02270.0060
119-0.1416-0.1097-0.0532
120-0.0241-0.0051-0.0498
121-0.0831-0.0880-0.0556
122-0.0651-0.0208-0.0130
123-0.0633-0.0329-0.0072
124-0.0730-0.0833-0.0516
1250.02880.01120.0811
126-0.0131-0.0251-0.0046
127-0.0625-0.0501-0.0532
128-0.0387-0.0438-0.0634
129-0.0840-0.0518-0.0549
1300.02800.00070.0203
131-0.0787-0.0577-0.0453
132-0.0495-0.0913-0.0548
133-0.0278-0.03010.0151
134-0.01700.01200.0318
135-0.0244-0.0025-0.0521
136-0.0183-0.0281-0.0126
137-0.00660.0088-0.0210
138-0.0708-0.0445-0.0456
1390.00940.01390.0108
1400.02620.03220.0028
1410.04880.02870.0203
142-0.01400.05460.0564
143-0.05100.00030.0397
1440.0161-0.0029-0.0450
145-0.0462-0.0499-0.0170
1460.06470.06210.0970
147-0.1056-0.1021-0.0460
1480.00600.0051-0.0217
149-0.02490.02990.1266
1500.00070.04190.0437
1510.04220.09270.0613
1520.0006-0.0294-0.0574
153-0.0691-0.0320-0.0815
154-0.0607-0.0706-0.0504
155-0.02180.0341-0.0520
156-0.0717-0.0454-0.0518
1570.14430.08470.1117
158-0.0501-0.0654-0.0644
1590.12140.09080.0734
160-0.0673-0.0146-0.0107
161-0.03360.0187-0.0216
162-0.0814-0.03150.0289
1630.03150.05270.0605
164-0.0515-0.0377-0.0454
1650.04070.06920.0574
1660.13020.17320.1837
167-0.1135-0.0777-0.0656
1680.01170.03550.0314
1690.08270.12140.0893
1700.03060.0518-0.0129
1710.04990.04380.0157
172-0.0374-0.05390.0714
173-0.0567-0.0687-0.0177
1740.09810.09650.1499
175-0.0189-0.0167-0.0742
176-0.0611-0.0238-0.0440
177-0.0308-0.0214-0.0426
178-0.02000.02050.0808
1790.06910.0260-0.0457
1800.03080.00980.0873
1810.05010.11530.1267
182-0.0464-0.0229-0.0447
1830.00020.03340.0367
184-0.0572-0.0258-0.0526
1850.03470.04940.0640
186-0.0146-0.0325-0.0606
1870.04290.17700.0856
1880.02070.0149-0.0486
1890.08310.10030.1165
1900.0367-0.00950.0675
191-0.0629-0.0607-0.0691
1920.13020.06830.1121
1930.07040.06020.0611
1940.0352-0.01290.0385
1950.14250.21090.1202
196-0.0211-0.0697-0.0502
1970.09990.11440.1029
198-0.0572-0.0711-0.0500
1990.03300.07480.0801
2000.11040.17320.1582
201-0.0347-0.0098-0.0158
202-0.0673-0.0315-0.0116
203-0.0809-0.0520-0.0414
204-0.0449-0.04850.0356
205-0.03610.01680.0370
206-0.0343-0.03130.0271
2070.03910.04590.0769
208-0.0889-0.0932-0.0536
209-0.0633-0.0857-0.0435
210-0.0018-0.0337-0.0673
211-0.0051-0.03270.0368
2120.00850.06500.0462
2130.02520.03770.0099
2140.12400.12990.0880
215-0.0550-0.0923-0.0435
2160.02230.0360-0.0435
2170.00400.00200.0347
2180.09590.04120.0662
2190.00470.02710.0033
220-0.0915-0.0941-0.0571
2210.01420.01240.0876
222-0.0787-0.0414-0.0606
2230.00000.03180.1058
2240.14340.07860.0389
225-0.0420-0.0581-0.0430
226-0.0585-0.02940.0670
227-0.02680.0025-0.0283
228-0.0611-0.03390.0488
229-0.0383-0.0315-0.0094
2300.01950.05770.0276
2310.10340.0513-0.0339
232-0.01350.03580.0444
233-0.0400-0.02140.0086
234-0.0020-0.0763-0.0439
2350.10120.08900.1328
236-0.0041-0.0289-0.0444
2370.10820.08240.0416
2380.0073-0.0038-0.0567
239-0.0032-0.0082-0.0535
240-0.00590.03390.0415
241-0.0893-0.03790.0195
242-0.0295-0.0128-0.0762
2430.05520.06310.1143
244-0.0413-0.0555-0.0550
2450.05670.0795-0.0128
2460.00780.01230.0089
2470.06640.05620.0576
248-0.0739-0.05620.0373
2490.0152-0.01010.0085
Source: Emolia / Emilia source archive · SHA-256 04491c3d8aff8226…
e07JapaneseEmolia / Emilia

Emolia JA · JA_B00000_S00310_W000001

Natural speech · Original speech · 13.2 s

だからそういう意味での資格のことを話したんだけど、まぁ、えっと、そう、今日資格なんだけど、私はね、こないだ、こないだ聴覚の話してた時に、女の人は聴覚がすごく、

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Contemplation
z +1.42 / +2.20raw 0.93 / 1.40
Contentment
z +1.79 / +1.53raw 0.92 / 0.79
Doubt
z +0.85 / +1.48raw 0.40 / 0.70
Concentration
z +1.16 / +1.28raw 1.49 / 1.58
Astonishment Surprise
z +0.57 / +0.94raw 0.33 / 0.54
Sourness
z +0.94 / +0.47raw 0.41 / 0.20
Interest
z +0.80 / +0.92raw 2.31 / 2.42

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Teacher or didactic style (S_TECH)
z +2.36 / +2.36raw 3.54 / 3.54
Vocal register (REGS)
z +1.88 / +2.08raw 3.58 / 3.84
Cartoonish style (S_CART)
z +1.96 / +1.31raw 4.30 / 3.39
Pitch range (RANG)
z +1.40 / +1.32raw 4.17 / 4.10
Playful style (S_PLAY)
z +1.36 / +1.32raw 4.14 / 4.09
Conversational style (S_CONV)
z +1.19 / +1.33raw 4.17 / 4.35
Tempo (TEMP)
z +0.97 / +1.16raw 3.27 / 3.49

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z +1.06 / +1.15raw 4.04 / 4.19
Vocal-burst blend (0–10)z +0.24 / +0.28raw 7.34 / 7.44
Content enjoyment (CE)z +0.03 / +0.36raw 4.83 / 5.17
Production quality (PQ)z -0.65 / -0.02raw 6.47 / 7.02
Speech signal quality, calibratedz +0.08 / +0.10raw 3.53 / 3.54
Background quality, calibratedz -0.05 / +0.24raw 3.34 / 3.61
Overall quality, calibratedz +0.09 / +0.24raw 2.89 / 2.99
P.808 MOS estimatez +0.11 / +0.28raw 3.47 / 3.56
score content enjoymentz +0.15 / -0.11raw 4.55 / 4.47
score overall qualityz +0.41 / +0.27raw 2.86 / 2.79
Background Noisez -0.23 / -0.15raw 0.75 / 0.77
Tempo (TEMP)z +0.97 / +1.16raw 3.27 / 3.49
Arousal (AROU)z -0.16 / -0.45raw 2.18 / 1.84
Valence (VALN)z +0.94 / +0.92raw 3.38 / 3.36
Recording quality (RCQL)z -0.04 / +0.10raw 2.72 / 2.85
Background noise (BKGN)z +0.42 / +0.42raw 3.02 / 3.02
talking speedz -1.35 / +0.10raw 3.94 / 9.42
Characters per second (CPS): 5.96 raw · z -0.82

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8290.850
Identity · 250D0.6200.605

Vocal-burst regions

Tiny
Base
0 s13.2 s

Verified burst intervals are unavailable for this clip.

Tiny: 2 regions, count-head estimate 1.23; Base: 1 regions, count-head estimate 1.11.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
Tiny10.020.360.34
Tiny24.004.200.20
Base14.004.220.22
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.368+0.630.226+0.380.0070.570
1Amusementemo_Amusement0.338+0.380.178+0.20-0.0010.885
2Angeremo_Anger0.196+0.31-0.084-0.140.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.333+0.570.544+0.940.0050.572
4Aweemo_Awe0.044+0.100.014+0.030.0000.427
5Bitternessemo_Bitterness0.052+0.090.010-0.010.0130.415
6Concentrationemo_Concentration1.492+1.161.584+1.280.6340.741
7Confusionemo_Confusion0.322+0.590.350+0.650.0010.540
8Contemplationemo_Contemplation0.929+1.421.398+2.200.0820.598
9Contemptemo_Contempt0.324+0.700.108+0.23-0.0000.464
10Contentmentemo_Contentment0.921+1.790.787+1.530.0000.514
11Disappointmentemo_Disappointment0.358+0.510.151+0.220.0000.700
12Disgustemo_Disgust0.176+0.340.019+0.020.0070.489
13Distressemo_Distress0.052+0.060.009+0.010.0000.821
14Doubtemo_Doubt0.402+0.850.697+1.480.0060.466
15Elationemo_Elation0.332+0.440.140+0.180.0120.728
16Embarrassmentemo_Embarrassment0.416+0.870.304+0.610.0430.430
17Emotional Numbnessemo_Emotional_Numbness-0.023-0.03-0.010-0.020.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.022-0.010.037+0.020.0280.539
19Fearemo_Fear-0.034-0.08-0.034-0.080.0150.610
20Helplessnessemo_Helplessness-0.014-0.02-0.042-0.05-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.474+0.470.687+0.750.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.677+0.750.151+0.000.1490.705
23Infatuationemo_Infatuation-0.009+0.01-0.042-0.05-0.0150.533
24Interestemo_Interest2.313+0.802.420+0.921.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness0.007+0.060.021+0.08-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.170+0.250.077+0.040.0590.442
27Longingemo_Longing0.229+0.230.323+0.370.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.071+0.030.078+0.040.0550.534
29Painemo_Pain-0.033-0.050.040+0.07-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.240+0.360.070+0.090.0170.614
31Prideemo_Pride0.127+0.250.097+0.190.0020.509
32Reliefemo_Relief0.088+0.080.007-0.040.0330.681
33Sadnessemo_Sadness0.032+0.04-0.001-0.000.0000.759
34Sexual Lustemo_Sexual_Lust0.139+0.220.025-0.010.0300.494
35Shameemo_Shame0.039+0.050.014+0.000.0130.510
36Sournessemo_Sourness0.407+0.940.201+0.47-0.0050.439
37Teasingemo_Teasing0.212+0.390.028+0.050.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.137+0.140.049+0.010.0410.697
39Triumphemo_Triumph-0.003+0.01-0.010-0.01-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg3.420+0.353.326+0.282.9841.238
41Arousal (AROU)vn_AROU_reg2.181-0.161.843-0.452.3591.147
42Arousal shift (ARSH)vn_ARSH_reg2.522-0.072.551-0.032.5800.867
43Vocal attack (ATCK)vn_ATCK_reg2.216+0.042.202+0.032.1621.295
44Background noise (BKGN)vn_BKGN_reg3.024+0.423.021+0.422.7030.761
45Brightness (BRGT)vn_BRGT_reg2.451+0.572.483+0.611.9910.813
46Speech chunking (CHNK)vn_CHNK_reg2.530+0.552.895+0.851.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg2.365+0.372.607+0.622.0180.945
48Cognitive load (COGL)vn_COGL_reg2.423+0.312.309+0.202.0941.061
49Dynamic arc (DARC)vn_DARC_reg1.646+0.021.317-0.221.6151.380
50Disfluency (DFLU)vn_DFLU_reg3.078+0.503.023+0.452.4881.184
51Emphasis (EMPH)vn_EMPH_reg3.362+0.503.313+0.462.6071.524
52Esthetics (ESTH)vn_ESTH_reg2.524+0.162.784+0.492.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.274+0.580.169+0.260.0860.322
54Focus (FOCS)vn_FOCS_reg2.317-0.022.314-0.032.3441.102
55Fullness (FULL)vn_FULL_reg1.609-0.191.831+0.061.7780.889
56Perceived gender (GEND)vn_GEND_reg1.605-1.541.629-1.534.0551.588
57Harmonicity (HARM)vn_HARM_reg1.721-0.911.873-0.732.4800.836
58Metallic character (METL)vn_METL_reg1.826+0.211.856+0.251.6210.958
59Pitch range (RANG)vn_RANG_reg4.173+1.404.096+1.322.9140.897
60Recording quality (RCQL)vn_RCQL_reg2.716-0.042.846+0.102.7540.888
61Vocal register (REGS)vn_REGS_reg3.582+1.883.837+2.081.1981.268
62Respiration (RESP)vn_RESP_reg3.820+1.083.386+0.672.6801.056
63Roughness (ROUG)vn_ROUG_reg2.777+0.842.829+0.902.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg2.222+0.102.348+0.252.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg2.896+0.912.890+0.902.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg2.502+0.592.674+0.761.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg1.591+0.182.037+0.721.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg3.807+0.973.345+0.532.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg2.301+0.162.550+0.502.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg2.044-0.032.238+0.212.0700.803
71Smoothness (SMTH)vn_SMTH_reg3.070+0.853.105+0.882.1431.093
72Stance (STNC)vn_STNC_reg2.948+0.282.870+0.212.6331.137
73Speech structure (STRU)vn_STRU_reg2.445+0.322.653+0.542.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg1.250-0.081.567+0.141.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg1.812+0.691.950+0.831.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg4.300+1.963.391+1.311.5601.401
77Casual style (S_CASU)vn_S_CASU_reg4.658+0.684.598+0.633.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg4.165+1.194.349+1.332.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg3.816+0.953.585+0.752.7211.148
80Formal style (S_FORM)vn_S_FORM_reg1.383+0.181.574+0.371.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg3.372+0.433.431+0.472.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg1.698+0.271.918+0.421.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.460+0.090.521+0.200.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg4.137+1.364.087+1.322.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg2.077+0.671.307+0.071.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg4.040+1.143.529+0.772.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg3.535+2.363.535+2.360.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg2.242+0.402.210+0.381.5971.618
89Tempo (TEMP)vn_TEMP_reg3.273+0.973.494+1.162.1171.190
90Tension (TENS)vn_TENS_reg1.590+0.041.250-0.361.5590.852
91Valence (VALN)vn_VALN_reg3.378+0.943.360+0.922.3281.118
92Valence shift (VALS)vn_VALS_reg3.064+0.513.347+0.772.5181.079
93Velocity flux (VFLX)vn_VFLX_reg2.500+0.202.596+0.282.2561.208
94Volatility (VOLT)vn_VOLT_reg3.115+0.603.046+0.522.5860.886
95Vulnerability (VULN)vn_VULN_reg2.434-0.222.330-0.322.6480.984
96Warmth (WARM)vn_WARM_reg3.160+0.643.210+0.692.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_64.043+1.064.194+1.152.3321.621
98Vocal-burst blend (0–10)blend_0_107.337+0.247.440+0.286.6762.738
99Legacy R_quality (no P3 test labels)R_quality1.408+0.931.442+1.031.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.245-0.023.256+0.023.2500.293
101Arousaleiv_extra:Arousal1.550+0.601.282-0.021.2890.435
102Authenticityeiv_extra:Authenticity2.884+1.422.856+1.232.6670.153
103Background Noiseeiv_extra:Background_Noise0.747-0.230.768-0.150.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.363+1.071.568+1.741.0400.303
105Gendereiv_extra:Gender1.527+2.151.356+1.98-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.694-0.531.666-0.651.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive2.660+1.712.635+1.651.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.362+0.662.288+0.432.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous1.260+0.971.304+1.100.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.753+1.480.807+1.650.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.521+0.450.244-0.410.3760.324
112Valenceeiv_extra:Valence0.307+0.790.120+0.300.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.306-0.931.323-0.901.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold0.283+0.770.344+0.970.0510.300
115score background qualityeiv_extra:score_background_quality3.440+0.203.362+0.093.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.553+0.154.474-0.114.5070.303
117score overall qualityeiv_extra:score_overall_quality2.858+0.412.787+0.272.6520.507
118score speech qualityeiv_extra:score_speech_quality1.765-0.291.760-0.321.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE4.829+0.035.167+0.364.7991.018
120Content usefulness (CU)audiobox:CU5.270-0.385.656+0.015.6491.001
121Production complexity (PC)audiobox:PC2.256-0.052.281-0.032.3311.617
122Production quality (PQ)audiobox:PQ6.473-0.657.024-0.027.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw3.973+0.113.977+0.123.9000.669
124Background quality, rawdnsmos:BAK_raw3.326+0.103.544+0.303.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.224+0.173.313+0.283.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.529+0.083.536+0.103.4880.501
127Background quality, calibrateddnsmos:BAK3.340-0.053.609+0.243.3890.928
128Overall quality, calibrateddnsmos:OVRL2.892+0.092.987+0.242.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.475+0.113.561+0.283.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.802+1.240.746+1.150.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection1.264+0.571.369+0.760.9470.554
132Agevoiceclap_attribute:Age3.104-0.233.051-0.323.2290.547
133Amusementvoiceclap_attribute:Amusement0.969+1.010.860+0.820.3920.572
134Angervoiceclap_attribute:Anger0.579-0.820.777-0.430.9920.506
135Arousalvoiceclap_attribute:Arousal1.748+0.531.773+0.571.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise1.077+1.190.958+0.940.5090.479
137Authenticityvoiceclap_attribute:Authenticity3.083+1.873.046+1.652.7700.167
138Awevoiceclap_attribute:Awe0.647+0.410.652+0.430.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.617+0.190.521-0.180.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.290-1.280.370-0.940.5920.236
141Concentrationvoiceclap_attribute:Concentration1.151-0.121.251+0.071.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.606+1.151.928+1.841.0690.468
143Confusionvoiceclap_attribute:Confusion1.187+1.631.321+2.030.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.896+0.291.065+0.700.7800.406
145Contemptvoiceclap_attribute:Contempt0.762+0.160.675-0.100.7090.339
146Contentmentvoiceclap_attribute:Contentment0.997+1.330.804+0.700.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.808+0.170.816+0.190.7600.291
148Disgustvoiceclap_attribute:Disgust0.544-0.390.595-0.140.6250.207
149Distressvoiceclap_attribute:Distress0.701-0.650.727-0.570.9270.350
150Doubtvoiceclap_attribute:Doubt0.651+0.620.805+1.430.5350.188
151Elationvoiceclap_attribute:Elation1.040+1.260.790+0.630.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.714+1.300.746+1.400.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.117-2.060.215-1.690.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.228-0.890.215-0.940.4620.261
155Fearvoiceclap_attribute:Fear0.437-0.850.467-0.710.6180.212
156Gendervoiceclap_attribute:Gender1.230+1.371.293+1.44-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.558-0.370.550-0.400.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.759-0.221.678-0.431.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism1.154+1.131.018+0.790.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability0.954-0.181.069+0.031.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.386-0.670.610+0.010.6080.332
162Interestvoiceclap_attribute:Interest2.148+1.252.120+1.191.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.156-0.750.281-0.480.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.783+1.060.873+1.430.5230.245
165Longingvoiceclap_attribute:Longing0.609+0.320.659+0.550.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.411-1.220.492-0.990.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive2.716+1.582.809+1.741.7750.594
168Painvoiceclap_attribute:Pain0.054-1.500.111-1.240.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.846+1.140.637+0.450.5030.301
170Pridevoiceclap_attribute:Pride1.213+1.240.913+0.380.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.459+0.442.462+0.452.3130.328
172Reliefvoiceclap_attribute:Relief1.476+1.091.224+0.481.0250.415
173Sadnessvoiceclap_attribute:Sadness0.660-0.290.621-0.420.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous1.271+0.821.372+1.000.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.297-0.810.346-0.690.6350.419
176Shamevoiceclap_attribute:Shame0.484-0.150.674+0.520.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.560+1.110.540+1.060.1600.360
178Sournessvoiceclap_attribute:Sourness0.742+0.120.712+0.020.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.584-0.110.481-0.400.6230.359
180Teasingvoiceclap_attribute:Teasing0.893+0.930.752+0.560.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.517+0.251.598+0.411.3840.529
182Triumphvoiceclap_attribute:Triumph0.883+0.460.704-0.030.7160.359
183Valencevoiceclap_attribute:Valence0.598+0.950.139+0.23-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached1.105-0.961.053-1.051.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold0.230+0.740.094+0.40-0.0660.400
186durationvoiceclap_attribute:duration14.171+0.5113.429+0.4010.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.347+0.203.378+0.273.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.827+1.174.802+0.994.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.772+0.332.746+0.272.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.853-0.541.889-0.021.8910.069
191talking speedvoiceclap_attribute:talking_speed3.941-1.359.418+0.109.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
0-0.0955-0.0788-0.0802
10.14640.13880.0945
20.12300.09010.1439
3-0.0806-0.0965-0.0817
4-0.0919-0.0577-0.0757
5-0.0336-0.0139-0.1020
60.06330.06410.1082
70.11810.07470.0903
80.04420.03830.0124
9-0.1082-0.0637-0.0693
10-0.0997-0.1041-0.0749
11-0.0905-0.1199-0.0942
12-0.0753-0.0486-0.0811
13-0.0675-0.05090.0071
140.04310.05540.0230
150.02080.0292-0.0166
16-0.1068-0.1003-0.0975
17-0.0905-0.0543-0.0539
18-0.0887-0.1154-0.0743
19-0.0474-0.0428-0.0757
200.14140.06330.1367
210.04150.0163-0.0741
220.04600.0027-0.0744
230.0082-0.0238-0.0562
240.18950.12290.1677
250.02320.07620.1375
26-0.0926-0.0875-0.0756
270.05440.06520.0030
28-0.0976-0.0622-0.0829
290.17400.16890.1281
300.05300.06330.0587
31-0.0962-0.0788-0.0744
320.10180.15230.0862
33-0.0757-0.1086-0.0888
34-0.0781-0.0690-0.0723
350.09690.13120.1117
36-0.0718-0.0383-0.0574
370.0048-0.06000.0090
38-0.0163-0.0292-0.0271
39-0.0100-0.0159-0.0388
40-0.0884-0.1161-0.0777
41-0.0605-0.0965-0.1011
42-0.1245-0.1063-0.0322
430.05590.0394-0.0017
44-0.1004-0.0890-0.0774
45-0.0292-0.0097-0.0659
460.09330.15910.1283
470.00180.0143-0.0392
48-0.1160-0.1071-0.0787
49-0.0806-0.0479-0.0814
50-0.0764-0.0897-0.0824
510.11380.07470.0416
520.0155-0.0375-0.0897
530.05090.02240.1393
54-0.0301-0.0194-0.0346
550.0133-0.0236-0.0350
56-0.0997-0.1033-0.0764
570.0348-0.0126-0.0514
58-0.03660.01610.0119
590.07570.13500.0615
60-0.0084-0.1003-0.0777
610.25880.28050.2373
620.00270.0087-0.0555
63-0.0849-0.1139-0.0765
640.03750.0349-0.0717
650.01120.04030.1062
660.08520.13650.1858
67-0.0817-0.0600-0.0777
680.16410.13350.1449
690.16120.17420.2391
700.0276-0.0202-0.0358
710.0316-0.0464-0.0557
720.09190.14100.0941
730.04070.03580.0295
74-0.1450-0.1124-0.0880
750.10110.09500.1926
760.18950.21120.1498
77-0.0919-0.0852-0.0736
78-0.1047-0.0988-0.0717
790.12800.05840.1215
80-0.0799-0.0833-0.0729
810.0527-0.00000.0427
82-0.0697-0.0660-0.0771
83-0.0548-0.0788-0.0374
84-0.0050-0.0358-0.0321
85-0.0887-0.0762-0.0856
860.10180.06150.0402
87-0.0247-0.0558-0.0575
88-0.0948-0.0879-0.0748
89-0.0615-0.01140.0029
900.04990.0226-0.0220
91-0.0142-0.00910.1083
92-0.05980.02870.0290
93-0.0249-0.0479-0.0655
940.0484-0.0201-0.0043
95-0.00530.0458-0.0644
960.0269-0.00560.0652
970.02740.01910.1445
98-0.0327-0.0102-0.0078
990.05300.09580.0371
1000.19090.20360.1652
101-0.0725-0.0253-0.0779
102-0.0771-0.0894-0.0862
103-0.1025-0.0867-0.0851
104-0.0170-0.00950.0208
1050.01820.06940.1146
1060.0079-0.0462-0.0077
107-0.0891-0.0769-0.0751
108-0.0940-0.0807-0.0425
1090.04140.04520.1183
110-0.0912-0.0867-0.0620
1110.09900.01400.1344
112-0.0725-0.0818-0.0207
113-0.1131-0.0618-0.1134
1140.04630.06330.0867
1150.0076-0.0206-0.0195
1160.20510.27600.1967
117-0.1011-0.1033-0.0938
118-0.0983-0.0826-0.0784
1190.08840.0603-0.0025
1200.04560.12740.0572
121-0.0679-0.0554-0.0042
122-0.1329-0.1274-0.0828
1230.04100.0447-0.0701
124-0.0672-0.0724-0.0052
125-0.0721-0.0995-0.0803
126-0.0460-0.0588-0.0695
1270.17540.08940.0649
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.0654-0.0413-0.0122
10.01250.0236-0.0368
2-0.1320-0.0896-0.0625
30.0491-0.00480.0326
4-0.0103-0.0362-0.0462
5-0.0098-0.0538-0.0531
6-0.0324-0.0431-0.0745
7-0.1215-0.0605-0.0515
80.11920.11120.0987
90.01420.06230.0000
10-0.0372-0.0527-0.0533
11-0.0257-0.0666-0.0019
120.13810.03550.0812
13-0.0788-0.0704-0.0448
14-0.0189-0.00790.0279
150.14640.12860.0799
16-0.0269-0.04250.0611
170.09040.02910.0677
180.08710.15830.0207
19-0.0399-0.0156-0.0132
20-0.0555-0.1053-0.0508
210.06910.03460.2435
220.08260.06050.0149
23-0.0577-0.0457-0.0491
240.08380.0160-0.0434
25-0.02830.0292-0.0010
26-0.0743-0.0191-0.0560
27-0.0638-0.0599-0.0413
280.07930.0495-0.0132
29-0.01440.0365-0.0086
300.10320.05590.1396
310.06790.01240.0327
320.0257-0.0319-0.0522
33-0.00330.0047-0.0106
340.07210.0016-0.0607
35-0.0243-0.0013-0.0199
360.0932-0.0050-0.0287
37-0.0290-0.0108-0.0351
38-0.0605-0.0681-0.0704
390.09260.04570.1366
400.0030-0.02370.0435
41-0.0342-0.0442-0.0465
420.07820.00710.0379
430.16080.14670.1741
440.0057-0.0402-0.0609
450.07600.08440.1195
46-0.03230.0056-0.0746
470.0507-0.04800.0334
480.0118-0.00450.0022
49-0.0649-0.1088-0.0540
50-0.0693-0.0422-0.0226
51-0.0663-0.1502-0.0515
520.0284-0.0202-0.0501
530.0156-0.0626-0.0491
540.06820.18390.0586
550.09430.07330.0455
560.03830.0378-0.0477
57-0.01480.04690.0081
58-0.03040.00750.0569
590.02700.0515-0.0143
60-0.02880.0345-0.0532
61-0.00790.0149-0.0017
62-0.0145-0.0074-0.0068
63-0.0538-0.0489-0.0552
640.09430.04390.0054
65-0.0291-0.01750.0019
660.0354-0.05500.0765
670.19630.10190.1194
680.0416-0.02210.0048
69-0.0577-0.0477-0.0651
70-0.0527-0.0684-0.0736
71-0.0530-0.0567-0.0532
720.04350.01670.0097
73-0.01730.02010.0233
74-0.0854-0.0937-0.0548
750.06410.0370-0.0620
760.00220.01860.0400
770.00000.0489-0.0046
780.02290.05150.0482
79-0.0013-0.0004-0.0543
80-0.0295-0.0908-0.0009
810.05190.04100.0356
820.0298-0.0710-0.0104
83-0.0449-0.1030-0.0451
840.11200.04070.0276
850.02810.05010.0342
86-0.0257-0.0223-0.0484
87-0.0183-0.0774-0.0433
88-0.03160.01640.0317
89-0.00590.0803-0.0490
90-0.01910.0005-0.0169
91-0.00060.08900.0191
92-0.0394-0.03750.0303
930.04940.0489-0.0391
94-0.0679-0.0439-0.0878
950.0037-0.0565-0.0263
96-0.0218-0.0579-0.0485
970.03630.0208-0.0530
98-0.01070.0495-0.0304
99-0.0876-0.0861-0.0679
1000.14530.14380.1656
101-0.0171-0.0949-0.0512
1020.04740.01090.0474
1030.16310.11170.1355
104-0.0679-0.0707-0.0275
105-0.0441-0.06810.0430
106-0.1159-0.0995-0.0418
107-0.0544-0.1019-0.0486
1080.02760.0365-0.0403
109-0.05380.0287-0.0521
1100.00590.02360.0789
1110.01980.0165-0.0447
1120.04220.04920.0635
1130.0165-0.02760.0205
1140.08540.07970.0294
1150.03290.12980.2183
116-0.00580.01240.0558
1170.10090.02910.1075
118-0.0632-0.0713-0.0707
1190.03720.03560.0068
120-0.0860-0.0178-0.0392
1210.16640.08030.1718
122-0.0599-0.0425-0.0512
123-0.0580-0.0614-0.0560
124-0.0052-0.06140.1215
125-0.0272-0.0434-0.0527
1260.04800.0004-0.0057
127-0.0191-0.0908-0.0558
128-0.0202-0.00640.0364
1290.0060-0.03620.0565
130-0.0298-0.0454-0.0545
131-0.0474-0.0176-0.0475
1320.08650.00350.0217
133-0.0030-0.0077-0.0536
134-0.0458-0.0704-0.0754
1350.10430.0174-0.0329
136-0.0092-0.0815-0.0581
137-0.0788-0.0815-0.0504
138-0.0438-0.0304-0.0478
1390.00370.12750.0539
140-0.0079-0.0215-0.0639
141-0.00090.0492-0.0536
142-0.0532-0.0591-0.0468
143-0.0452-0.06430.0406
144-0.0588-0.01190.0887
1450.04630.00660.0010
1460.11150.00530.0717
147-0.0721-0.0931-0.0482
148-0.0117-0.0051-0.0598
149-0.03940.0002-0.0698
150-0.0849-0.0599-0.0534
1510.01650.05410.0942
152-0.0030-0.0259-0.0345
1530.02330.0387-0.0692
154-0.0156-0.0573-0.0529
1550.00330.00620.0388
156-0.0483-0.0559-0.0543
1570.10820.09490.0460
1580.09040.06550.0622
159-0.0257-0.05030.0523
1600.0258-0.0608-0.0515
1610.06070.10240.1003
162-0.0177-0.0978-0.0629
163-0.0385-0.0303-0.0281
164-0.0288-0.0608-0.0476
1650.07990.16990.0855
166-0.0455-0.0856-0.0630
167-0.0507-0.0483-0.0687
1680.01770.02660.0332
169-0.0616-0.0762-0.0619
1700.04550.01330.0614
1710.05100.00480.0602
172-0.0915-0.1455-0.0628
173-0.0164-0.0365-0.0635
1740.0505-0.00860.0070
175-0.1137-0.0937-0.0660
176-0.0505-0.0535-0.0461
1770.02520.02100.1116
1780.03290.0068-0.0055
1790.02810.0266-0.0256
1800.0546-0.00110.0445
181-0.0349-0.0345-0.0773
182-0.0326-0.0052-0.0285
183-0.0079-0.03900.0270
1840.10210.0133-0.0551
1850.07820.05500.0044
1860.03660.04630.0562
187-0.04830.0378-0.0313
1880.04600.01740.0188
1890.09710.07130.1033
1900.0474-0.0112-0.0654
1910.04960.04710.0227
192-0.0316-0.1013-0.0656
193-0.1270-0.0925-0.0571
1940.03510.0227-0.0703
195-0.0715-0.0214-0.0515
196-0.02520.0064-0.0526
197-0.0657-0.0821-0.0758
198-0.0638-0.0687-0.0524
1990.0358-0.05590.0242
2000.0123-0.02560.0448
2010.01860.0428-0.0474
202-0.1093-0.0585-0.0685
203-0.0483-0.01800.0116
204-0.04020.0375-0.0381
2050.0063-0.0483-0.0472
2060.18190.07570.0410
207-0.0117-0.0195-0.0446
208-0.0749-0.0498-0.0562
2090.03080.0230-0.0873
210-0.0113-0.02750.0194
211-0.0771-0.0856-0.0417
2120.0164-0.0457-0.0509
213-0.0008-0.01850.0301
2140.03260.1065-0.0432
215-0.0139-0.0160-0.0456
216-0.01270.01510.0113
2170.02870.02440.1086
2180.02980.0165-0.0629
2190.0388-0.01750.0323
220-0.0860-0.1234-0.0598
221-0.0943-0.0075-0.0470
222-0.0363-0.02120.0004
223-0.0202-0.0579-0.0435
2240.05410.0368-0.0009
225-0.0632-0.0180-0.0451
2260.00920.0448-0.0403
2270.08260.10530.2041
228-0.0262-0.0704-0.0877
229-0.0776-0.0873-0.0594
230-0.0496-0.0661-0.0552
2310.11760.06050.0351
2320.05600.05620.0312
233-0.0596-0.0803-0.0067
2340.11430.01150.0224
2350.06020.1176-0.0445
236-0.0460-0.0451-0.0197
237-0.01200.04630.0793
238-0.0904-0.0434-0.0594
2390.0079-0.0434-0.0247
240-0.0921-0.1158-0.0553
2410.0849-0.0071-0.0322
242-0.0361-0.0098-0.0723
243-0.0113-0.0704-0.0696
2440.0068-0.01650.0372
2450.19190.28290.1749
246-0.0574-0.0960-0.0439
2470.02440.01790.0804
248-0.0444-0.0570-0.0620
2490.10320.02600.1345
Source: Emolia / Emilia source archive · SHA-256 028dbeae0f1af297…
e08JapaneseEmolia / Emilia

Emolia JA · JA_B00000_S00080_W000086

Natural speech · Original speech · 3.2 s

午後になったらうなぎを取りに行ってぇ。

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Impatience and Irritability
z +1.59 / +0.67raw 1.27 / 0.62
Confusion
z +1.15 / +0.79raw 0.62 / 0.43
Doubt
z +0.86 / +0.51raw 0.41 / 0.24
Anger
z +0.79 / +0.29raw 0.49 / 0.18
Longing
z +0.73 / +0.57raw 0.55 / 0.45
Disappointment
z +0.27 / +0.48raw 0.19 / 0.33
Teasing
z +0.45 / +0.03raw 0.25 / 0.02

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Vocal register (REGS)
z +2.33 / +2.42raw 4.15 / 4.27
Pitch range (RANG)
z +1.57 / +1.68raw 4.33 / 4.42
Brightness (BRGT)
z +1.63 / +1.42raw 3.32 / 3.15
Storytelling style (S_STRY)
z +1.57 / +1.32raw 4.63 / 4.29
Conversational style (S_CONV)
z +1.52 / +1.37raw 4.60 / 4.40
Nasal resonance (R_NASL)
z +1.51 / +1.33raw 4.37 / 4.19
Head resonance (R_HEAD)
z +1.32 / +1.39raw 3.24 / 3.29

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z +1.16 / +1.21raw 4.21 / 4.29
Vocal-burst blend (0–10)z -0.70 / -0.48raw 4.76 / 5.36
Content enjoyment (CE)z -0.04 / +0.04raw 4.76 / 4.84
Production quality (PQ)z -0.43 / -0.30raw 6.67 / 6.78
Speech signal quality, calibratedz -0.51 / -0.32raw 3.24 / 3.33
Background quality, calibratedz +0.28 / +0.26raw 3.65 / 3.63
Overall quality, calibratedz -0.02 / -0.05raw 2.83 / 2.81
P.808 MOS estimatez -0.26 / -0.44raw 3.29 / 3.20
score content enjoymentz -0.98 / -1.09raw 4.21 / 4.18
score overall qualityz -0.17 / -0.29raw 2.56 / 2.50
Background Noisez +0.12 / +0.04raw 0.84 / 0.82
Tempo (TEMP)z +0.11 / +0.09raw 2.24 / 2.23
Arousal (AROU)z -0.05 / -0.07raw 2.31 / 2.28
Valence (VALN)z +0.51 / +0.31raw 2.90 / 2.67
Recording quality (RCQL)z -0.46 / -0.83raw 2.35 / 2.01
Background noise (BKGN)z -0.32 / -0.41raw 2.46 / 2.39
talking speedz +0.40 / +0.64raw 10.51 / 11.45
Characters per second (CPS): 5.86 raw · z -0.84

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.7310.754
Identity · 250D0.7170.757

Vocal-burst regions

Tiny
Base
0 s3.2 s

Verified burst intervals are unavailable for this clip.

Tiny: 0 regions, count-head estimate 0.06; Base: 0 regions, count-head estimate 0.00.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.131+0.220.122+0.200.0070.570
1Amusementemo_Amusement0.244+0.280.312+0.35-0.0010.885
2Angeremo_Anger0.487+0.790.182+0.290.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.066+0.110.043+0.070.0050.572
4Aweemo_Awe0.007+0.02-0.011-0.020.0000.427
5Bitternessemo_Bitterness0.025+0.030.037+0.060.0130.415
6Concentrationemo_Concentration0.664+0.040.766+0.180.6340.741
7Confusionemo_Confusion0.622+1.150.426+0.790.0010.540
8Contemplationemo_Contemplation0.064-0.030.103+0.030.0820.598
9Contemptemo_Contempt0.111+0.240.018+0.04-0.0000.464
10Contentmentemo_Contentment0.012+0.020.085+0.160.0000.514
11Disappointmentemo_Disappointment0.191+0.270.333+0.480.0000.700
12Disgustemo_Disgust-0.016-0.05-0.005-0.030.0070.489
13Distressemo_Distress0.168+0.200.219+0.270.0000.821
14Doubtemo_Doubt0.409+0.860.242+0.510.0060.466
15Elationemo_Elation0.031+0.030.073+0.080.0120.728
16Embarrassmentemo_Embarrassment0.062+0.040.060+0.040.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.054+0.050.182+0.190.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.069+0.080.051+0.040.0280.539
19Fearemo_Fear0.066+0.080.029+0.020.0150.610
20Helplessnessemo_Helplessness0.224+0.330.265+0.39-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.095-0.030.077-0.050.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability1.273+1.590.624+0.670.1490.705
23Infatuationemo_Infatuation-0.032-0.03-0.047-0.06-0.0150.533
24Interestemo_Interest1.028-0.620.985-0.671.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness0.049+0.13-0.009+0.03-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.027-0.070.019-0.090.0590.442
27Longingemo_Longing0.546+0.730.445+0.570.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.054-0.000.050-0.010.0550.534
29Painemo_Pain-0.021-0.030.031+0.06-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.032+0.020.073+0.090.0170.614
31Prideemo_Pride-0.011-0.030.053+0.100.0020.509
32Reliefemo_Relief0.051+0.030.062+0.040.0330.681
33Sadnessemo_Sadness0.204+0.270.317+0.420.0000.759
34Sexual Lustemo_Sexual_Lust0.018-0.02-0.006-0.070.0300.494
35Shameemo_Shame-0.042-0.11-0.007-0.040.0130.510
36Sournessemo_Sourness0.007+0.03-0.016-0.02-0.0050.439
37Teasingemo_Teasing0.248+0.450.018+0.030.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude-0.013-0.080.008-0.050.0410.697
39Triumphemo_Triumph-0.016-0.020.024+0.06-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg0.358-2.120.191-2.262.9841.238
41Arousal (AROU)vn_AROU_reg2.306-0.052.279-0.072.3591.147
42Arousal shift (ARSH)vn_ARSH_reg3.300+0.833.187+0.702.5800.867
43Vocal attack (ATCK)vn_ATCK_reg2.337+0.142.093-0.052.1621.295
44Background noise (BKGN)vn_BKGN_reg2.460-0.322.392-0.412.7030.761
45Brightness (BRGT)vn_BRGT_reg3.319+1.633.148+1.421.9910.813
46Speech chunking (CHNK)vn_CHNK_reg1.432-0.341.505-0.281.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg1.170-0.901.050-1.022.0180.945
48Cognitive load (COGL)vn_COGL_reg1.965-0.121.694-0.382.0941.061
49Dynamic arc (DARC)vn_DARC_reg2.592+0.712.339+0.521.6151.380
50Disfluency (DFLU)vn_DFLU_reg2.917+0.363.184+0.592.4881.184
51Emphasis (EMPH)vn_EMPH_reg2.356-0.161.938-0.442.6071.524
52Esthetics (ESTH)vn_ESTH_reg2.915+0.662.885+0.622.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.024-0.190.009-0.240.0860.322
54Focus (FOCS)vn_FOCS_reg1.923-0.381.970-0.342.3441.102
55Fullness (FULL)vn_FULL_reg0.757-1.150.826-1.071.7780.889
56Perceived gender (GEND)vn_GEND_reg-0.100-2.62-0.199-2.684.0551.588
57Harmonicity (HARM)vn_HARM_reg2.973+0.593.041+0.672.4800.836
58Metallic character (METL)vn_METL_reg2.122+0.521.996+0.391.6210.958
59Pitch range (RANG)vn_RANG_reg4.326+1.574.424+1.682.9140.897
60Recording quality (RCQL)vn_RCQL_reg2.345-0.462.013-0.832.7540.888
61Vocal register (REGS)vn_REGS_reg4.151+2.334.269+2.421.1981.268
62Respiration (RESP)vn_RESP_reg2.925+0.233.285+0.572.6801.056
63Roughness (ROUG)vn_ROUG_reg1.773-0.291.873-0.182.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg1.341-0.941.172-1.142.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg3.237+1.323.289+1.392.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg3.218+1.303.026+1.111.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg0.657-0.930.696-0.881.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg4.368+1.514.186+1.332.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg3.172+1.342.905+0.972.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg2.556+0.602.427+0.442.0700.803
71Smoothness (SMTH)vn_SMTH_reg1.810-0.301.707-0.402.1431.093
72Stance (STNC)vn_STNC_reg2.382-0.222.254-0.332.6331.137
73Speech structure (STRU)vn_STRU_reg1.676-0.481.536-0.632.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg0.305-0.750.057-0.931.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg0.684-0.390.473-0.591.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg3.209+1.183.134+1.121.5601.401
77Casual style (S_CASU)vn_S_CASU_reg5.206+1.115.325+1.203.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg4.604+1.524.400+1.372.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg2.912+0.172.825+0.092.7211.148
80Formal style (S_FORM)vn_S_FORM_reg0.623-0.560.466-0.711.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg1.284-1.201.294-1.192.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg0.769-0.370.968-0.231.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.411+0.010.270-0.240.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg3.751+1.063.721+1.042.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg-0.171-1.07-0.041-0.971.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg4.628+1.574.291+1.322.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg0.375-0.220.112-0.440.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg1.645+0.031.705+0.071.5971.618
89Tempo (TEMP)vn_TEMP_reg2.242+0.112.229+0.092.1171.190
90Tension (TENS)vn_TENS_reg1.531-0.031.470-0.101.5590.852
91Valence (VALN)vn_VALN_reg2.897+0.512.674+0.312.3281.118
92Valence shift (VALS)vn_VALS_reg3.043+0.492.569+0.052.5181.079
93Velocity flux (VFLX)vn_VFLX_reg3.315+0.883.098+0.702.2561.208
94Volatility (VOLT)vn_VOLT_reg3.166+0.653.183+0.672.5860.886
95Vulnerability (VULN)vn_VULN_reg3.189+0.553.363+0.732.6480.984
96Warmth (WARM)vn_WARM_reg1.665-1.001.573-1.112.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_64.206+1.164.294+1.212.3321.621
98Vocal-burst blend (0–10)blend_0_104.765-0.705.365-0.486.6762.738
99Legacy R_quality (no P3 test labels)R_quality1.185+0.271.250+0.461.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age2.780-1.601.614-5.583.2500.293
101Arousaleiv_extra:Arousal1.357+0.161.238-0.121.2890.435
102Authenticityeiv_extra:Authenticity2.715+0.312.759+0.602.6670.153
103Background Noiseeiv_extra:Background_Noise0.838+0.120.818+0.040.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.886+2.801.829+2.611.0400.303
105Gendereiv_extra:Gender0.811+1.430.627+1.25-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.571-1.021.520-1.231.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive2.039+0.381.939+0.171.8590.470
108Recording Qualityeiv_extra:Recording_Quality1.762-1.151.720-1.282.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous1.053+0.361.036+0.310.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.730+1.400.922+2.010.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.204-0.530.005-1.140.3760.324
112Valenceeiv_extra:Valence0.020+0.04-0.050-0.150.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.248-1.071.228-1.111.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold0.133+0.270.145+0.310.0510.300
115score background qualityeiv_extra:score_background_quality3.319+0.033.300-0.003.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.209-0.984.176-1.094.5070.303
117score overall qualityeiv_extra:score_overall_quality2.565-0.172.504-0.292.6520.507
118score speech qualityeiv_extra:score_speech_quality1.909+0.691.932+0.861.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE4.760-0.044.841+0.044.7991.018
120Content usefulness (CU)audiobox:CU5.321-0.335.525-0.125.6491.001
121Production complexity (PC)audiobox:PC1.820-0.321.700-0.392.3311.617
122Production quality (PQ)audiobox:PQ6.665-0.436.781-0.307.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw3.495-0.613.615-0.433.9000.669
124Background quality, rawdnsmos:BAK_raw3.680+0.423.599+0.353.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.106+0.033.050-0.043.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.235-0.513.327-0.323.4880.501
127Background quality, calibrateddnsmos:BAK3.647+0.283.633+0.263.3890.928
128Overall quality, calibrateddnsmos:OVRL2.826-0.022.809-0.052.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.291-0.263.204-0.443.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.056+0.090.002+0.000.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection1.270+0.581.125+0.320.9470.554
132Agevoiceclap_attribute:Age1.331-3.470.796-4.453.2290.547
133Amusementvoiceclap_attribute:Amusement0.946+0.970.829+0.760.3920.572
134Angervoiceclap_attribute:Anger0.236-1.490.518-0.940.9920.506
135Arousalvoiceclap_attribute:Arousal1.657+0.381.732+0.501.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise1.100+1.230.820+0.650.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.946+1.052.850+0.482.7700.167
138Awevoiceclap_attribute:Awe0.319-0.850.335-0.780.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.533-0.130.562-0.020.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.212-1.620.235-1.520.5920.236
141Concentrationvoiceclap_attribute:Concentration0.360-1.610.448-1.441.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.760+1.482.005+2.001.0690.468
143Confusionvoiceclap_attribute:Confusion0.827+0.550.860+0.650.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.028-1.860.151-1.550.7800.406
145Contemptvoiceclap_attribute:Contempt0.584-0.370.546-0.480.7090.339
146Contentmentvoiceclap_attribute:Contentment0.684+0.310.561-0.090.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.320-1.510.472-0.990.7600.291
148Disgustvoiceclap_attribute:Disgust0.386-1.150.398-1.090.6250.207
149Distressvoiceclap_attribute:Distress0.445-1.380.634-0.840.9270.350
150Doubtvoiceclap_attribute:Doubt0.570+0.190.684+0.800.5350.188
151Elationvoiceclap_attribute:Elation0.751+0.530.678+0.350.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.548+0.770.385+0.250.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.436-0.860.519-0.540.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.361-0.390.299-0.620.4620.261
155Fearvoiceclap_attribute:Fear0.421-0.930.488-0.610.6180.212
156Gendervoiceclap_attribute:Gender0.850+1.000.956+1.10-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.560-0.370.601-0.210.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.434-1.081.426-1.111.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism1.004+0.750.919+0.540.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability0.870-0.340.921-0.241.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.756+0.450.833+0.680.6080.332
162Interestvoiceclap_attribute:Interest1.346-0.571.514-0.191.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.535+0.070.414-0.190.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.520-0.010.689+0.680.5230.245
165Longingvoiceclap_attribute:Longing0.634+0.430.753+0.980.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.408-1.230.398-1.260.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive2.203+0.722.313+0.911.7750.594
168Painvoiceclap_attribute:Pain0.128-1.160.216-0.760.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.719+0.720.666+0.540.5030.301
170Pridevoiceclap_attribute:Pride0.174-1.740.152-1.800.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.064-0.761.895-1.282.3130.328
172Reliefvoiceclap_attribute:Relief0.805-0.530.605-1.011.0250.415
173Sadnessvoiceclap_attribute:Sadness0.591-0.510.802+0.160.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous1.646+1.481.543+1.300.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.412-0.530.433-0.480.6350.419
176Shamevoiceclap_attribute:Shame0.368-0.560.294-0.820.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.970+2.250.925+2.130.1600.360
178Sournessvoiceclap_attribute:Sourness0.488-0.750.537-0.580.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.222-1.120.185-1.220.6230.359
180Teasingvoiceclap_attribute:Teasing0.928+1.020.817+0.730.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude0.988-0.750.560-1.561.3840.529
182Triumphvoiceclap_attribute:Triumph0.202-1.430.117-1.670.7160.359
183Valencevoiceclap_attribute:Valence0.390+0.620.395+0.63-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached0.988-1.171.036-1.081.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold0.773+2.100.658+1.81-0.0660.400
186durationvoiceclap_attribute:duration3.867-1.113.322-1.1910.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.288+0.093.321+0.153.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.550-0.724.537-0.804.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.419-0.522.434-0.492.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality2.040+2.152.057+2.401.8910.069
191talking speedvoiceclap_attribute:talking_speed10.515+0.4011.446+0.649.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.13160.0712-0.0814
10.07290.0402-0.0540
2-0.0625-0.0459-0.0003
3-0.0436-0.0662-0.0829
4-0.0804-0.0810-0.0768
5-0.1107-0.0885-0.1035
60.04720.08070.0227
70.18240.18230.1780
80.03920.04070.1646
90.0007-0.0453-0.0027
10-0.0742-0.0679-0.0760
11-0.0964-0.0971-0.0066
12-0.0814-0.0744-0.0823
13-0.0076-0.02800.0695
140.04620.0121-0.0120
150.03710.0639-0.0064
16-0.0886-0.0741-0.0069
17-0.0798-0.0872-0.0850
18-0.0791-0.0676-0.0754
19-0.0912-0.0590-0.0773
20-0.01680.01610.0549
21-0.0303-0.0892-0.0752
220.0049-0.0512-0.0769
23-0.0931-0.10760.0070
240.02060.0233-0.0169
250.03320.05050.1494
26-0.0729-0.0718-0.0767
27-0.0632-0.0374-0.0763
28-0.0873-0.0977-0.0901
29-0.0485-0.0443-0.0614
30-0.01070.04980.0687
31-0.0321-0.0449-0.0191
320.05050.08530.0403
33-0.0811-0.1017-0.0901
34-0.0938-0.0741-0.0733
350.10360.1207-0.0025
36-0.0983-0.1128-0.0879
37-0.0970-0.0931-0.0969
380.13550.12070.1116
39-0.0905-0.0741-0.0825
400.12500.10100.1128
41-0.0254-0.0639-0.1078
42-0.0899-0.1023-0.0825
430.0100-0.00650.1125
440.07360.06430.0736
450.14330.11280.1458
46-0.02410.0036-0.0056
47-0.0697-0.07580.0143
48-0.0746-0.0721-0.0798
490.14460.11410.0622
500.0313-0.0298-0.0099
510.10940.13510.1143
52-0.0557-0.05510.0665
53-0.0329-0.03080.0363
54-0.0951-0.1089-0.0915
550.07420.0279-0.0267
56-0.0847-0.0820-0.0776
57-0.0309-0.0630-0.0897
580.14070.14690.1584
590.07130.10300.0900
600.20840.16000.1273
610.15440.14360.0806
620.07780.08100.0964
630.09440.11670.1606
640.05310.03410.0947
65-0.0729-0.0685-0.0667
660.15830.21910.1520
67-0.0762-0.0738-0.0789
68-0.0824-0.02390.0052
69-0.0593-0.0872-0.0294
700.10420.13770.0362
710.23580.23220.1719
72-0.0918-0.0958-0.0792
730.01650.0014-0.1000
740.0019-0.02740.0784
750.12240.11280.0221
760.09440.08170.0197
77-0.0795-0.0761-0.0747
78-0.0736-0.0603-0.0728
79-0.0521-0.0346-0.0709
80-0.0677-0.0656-0.0740
810.16610.11740.1353
82-0.0847-0.0833-0.0782
830.13290.13120.0688
840.0056-0.00090.0223
85-0.0075-0.0408-0.0869
860.05110.0305-0.0525
87-0.0990-0.0905-0.0944
88-0.0257-0.0067-0.0584
89-0.0752-0.0892-0.0670
900.11070.15410.0513
91-0.0368-0.0354-0.0752
92-0.0765-0.0738-0.0794
930.06550.04560.1061
940.02690.0659-0.0127
95-0.0089-0.0479-0.0561
96-0.0655-0.07800.0940
970.21100.22170.2805
98-0.0847-0.0931-0.0729
990.09570.0790-0.0687
1000.15440.20990.0232
101-0.0821-0.0872-0.0790
102-0.00230.02000.0577
103-0.1009-0.1043-0.0863
104-0.00770.00650.0314
105-0.0977-0.0679-0.0789
106-0.0321-0.00710.0082
107-0.0778-0.0682-0.0762
108-0.0892-0.0964-0.0887
1090.06510.08230.0091
1100.06810.06820.1852
1110.03090.01550.1504
112-0.1055-0.0846-0.0806
113-0.0661-0.0777-0.1081
1140.04360.0185-0.0236
1150.10940.06530.0565
1160.15170.12260.0958
1170.01180.05380.1322
118-0.0752-0.0728-0.0796
119-0.00680.0004-0.0403
1200.04490.0323-0.0841
1210.14260.12860.1678
122-0.0602-0.05670.0586
1230.01740.01000.1064
1240.06550.0502-0.0390
125-0.01140.0234-0.0815
126-0.0306-0.0167-0.0427
1270.18240.19020.1580
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
00.10970.11950.1185
1-0.0661-0.0432-0.0423
2-0.0476-0.0798-0.0538
3-0.0031-0.0300-0.0303
40.00980.0453-0.0398
50.14520.13140.0320
60.07260.04780.0358
7-0.0071-0.0112-0.0443
8-0.01110.03070.0698
90.19920.18570.1930
100.04390.06440.0098
11-0.0867-0.1054-0.0591
120.07320.06410.0302
130.25440.23790.0681
14-0.0625-0.0334-0.0624
15-0.0153-0.0085-0.0515
160.08610.03750.1017
17-0.0974-0.0722-0.0525
180.07260.0231-0.0202
19-0.0839-0.0722-0.0086
200.03970.04050.0189
21-0.0509-0.0394-0.0405
22-0.03250.0103-0.0453
23-0.0675-0.0831-0.0423
24-0.0332-0.0451-0.0373
250.03390.05110.0369
260.05710.04020.0483
27-0.0633-0.0383-0.0355
28-0.0155-0.0147-0.0415
29-0.0450-0.0119-0.0421
300.01680.05810.0005
310.13280.13470.0487
320.15310.14230.1063
33-0.0473-0.0625-0.0392
340.16770.11790.0688
35-0.0371-0.0424-0.0563
360.06220.0075-0.0462
37-0.0134-0.00710.0051
38-0.0021-0.0101-0.0318
39-0.0805-0.0549-0.0651
40-0.0405-0.0182-0.0506
41-0.0805-0.0538-0.0502
420.0249-0.0139-0.0434
430.02280.04940.0138
440.0017-0.0367-0.0526
45-0.01190.04320.0467
46-0.0793-0.0744-0.0642
470.09740.00360.0531
480.09340.09990.1206
49-0.0495-0.0364-0.0423
500.03490.0367-0.0609
510.11140.03050.0458
520.04020.0633-0.0183
53-0.0726-0.0771-0.0423
54-0.02080.00370.0203
550.04590.05510.0130
56-0.0546-0.0258-0.0601
570.04000.0633-0.0035
58-0.0495-0.0228-0.0420
59-0.02270.0037-0.0609
60-0.0574-0.0733-0.0458
61-0.0523-0.0303-0.0406
62-0.0263-0.0853-0.0370
630.04670.02580.0268
64-0.0777-0.0809-0.0487
650.05370.10480.1407
660.09900.0798-0.0230
67-0.0360-0.0831-0.0211
680.09170.10700.0577
690.02980.0248-0.0004
700.0055-0.0115-0.0634
71-0.0310-0.0426-0.0458
720.00690.02240.0077
730.04560.06060.0011
74-0.0929-0.0956-0.0472
75-0.0105-0.0126-0.0172
760.03520.03000.0008
770.00690.01510.1444
780.03660.09720.0795
79-0.0647-0.0749-0.0468
800.05010.09940.1206
81-0.0380-0.00230.0036
820.03290.0185-0.0420
830.04810.05130.0300
84-0.00820.01720.0188
85-0.0297-0.0289-0.0467
86-0.0227-0.0492-0.0417
87-0.0591-0.0633-0.0372
880.13840.11620.0723
89-0.0644-0.0383-0.0422
900.0339-0.0388-0.0462
910.11870.08200.0912
92-0.0098-0.0549-0.0403
930.09170.12490.1135
940.00460.0115-0.0367
950.0242-0.02670.0130
96-0.0347-0.0229-0.0417
97-0.02480.02380.0326
98-0.0200-0.0494-0.0493
990.00260.01850.0525
1000.0478-0.0154-0.0526
101-0.0602-0.0614-0.0441
102-0.0442-0.03180.0233
103-0.0568-0.0227-0.0559
104-0.1300-0.1249-0.0608
105-0.0900-0.0535-0.0607
1060.07990.08580.1333
107-0.0428-0.0581-0.0418
108-0.0228-0.0551-0.0347
109-0.0591-0.0329-0.0449
110-0.0414-0.03580.0081
111-0.0262-0.0394-0.0385
112-0.0613-0.0380-0.0377
113-0.1024-0.1206-0.0796
114-0.1097-0.0853-0.0651
1150.02030.02340.1199
1160.01830.02620.0386
117-0.0305-0.0388-0.0548
1180.04330.07550.0628
1190.13900.09340.1584
120-0.0183-0.0137-0.0289
1210.07820.06300.1125
122-0.0450-0.0394-0.0441
123-0.0170-0.0074-0.0248
1240.04730.07220.0919
1250.17670.18250.1885
1260.06700.08800.0400
1270.03600.00260.0318
128-0.0633-0.03230.0083
1290.11030.13310.1531
1300.04590.06870.1849
131-0.0257-0.0684-0.0408
1320.05180.13140.0492
1330.02910.0080-0.0443
1340.0777-0.0293-0.0642
1350.02930.02860.0830
136-0.0549-0.0595-0.0500
137-0.0447-0.0443-0.0433
138-0.0532-0.0472-0.0412
139-0.0425-0.0798-0.0477
1400.01030.00990.0228
1410.02810.04130.0864
142-0.0248-0.0581-0.0238
143-0.0124-0.0267-0.0280
144-0.0026-0.03640.0718
145-0.0805-0.0443-0.0577
1460.04900.09830.0882
147-0.0929-0.0831-0.0415
148-0.0692-0.0825-0.0515
149-0.0175-0.04780.0067
150-0.0056-0.0221-0.0205
151-0.0895-0.0820-0.0532
152-0.0506-0.0472-0.0518
153-0.0591-0.0673-0.0660
1540.0308-0.06250.0069
155-0.03020.0025-0.0609
1560.02860.02080.0338
1570.04670.0706-0.0150
158-0.0270-0.0136-0.0067
1590.07320.05000.0893
160-0.0001-0.0573-0.0443
1610.07710.11080.0349
1620.00330.01960.0382
163-0.0394-0.0208-0.0343
164-0.0506-0.0261-0.0410
1650.05660.11890.0914
166-0.0346-0.0432-0.0543
167-0.0416-0.0097-0.0591
1680.02360.02320.0560
1690.0028-0.0652-0.0020
170-0.0568-0.0809-0.0437
1710.05540.04100.0149
1720.03490.0333-0.0311
173-0.0419-0.0111-0.0547
1740.03190.02240.0132
1750.08780.05780.0133
176-0.0605-0.0573-0.0397
177-0.0551-0.0644-0.0385
1780.06980.0581-0.0137
179-0.0219-0.0013-0.0351
1800.04840.04670.0702
1810.0184-0.0464-0.0572
182-0.0349-0.00470.0463
183-0.0193-0.08850.0023
184-0.0176-0.0486-0.0474
1850.03600.02670.0211
186-0.0137-0.0086-0.0365
1870.04330.10430.1133
1880.00230.0383-0.0096
189-0.00830.02060.0284
190-0.0509-0.0346-0.0813
191-0.0706-0.0592-0.0623
192-0.00420.0223-0.0565
193-0.0535-0.0540-0.0492
1940.06780.08530.0450
195-0.0239-0.0319-0.0443
196-0.0481-0.0234-0.0453
1970.10520.07170.0470
198-0.0644-0.0842-0.0451
1990.17450.13250.2595
200-0.03400.0115-0.0068
201-0.0137-0.0156-0.0439
2020.02380.0085-0.0412
2030.06890.06790.1236
204-0.0149-0.0204-0.0185
205-0.0689-0.0472-0.0812
2060.05990.0513-0.0481
207-0.0833-0.0521-0.0630
208-0.0402-0.0262-0.0484
209-0.0642-0.0329-0.0751
210-0.0793-0.0809-0.0607
2110.09000.0815-0.0146
212-0.0867-0.0310-0.0609
2130.03910.0760-0.0300
214-0.0726-0.0155-0.0092
2150.00840.0090-0.0026
2160.08330.05430.0319
2170.0013-0.0098-0.0631
218-0.0201-0.0535-0.0541
2190.01540.03160.0820
220-0.0591-0.1016-0.0515
2210.03460.01900.0093
2220.13680.10320.0729
2230.0347-0.0303-0.0374
224-0.0675-0.0003-0.0426
225-0.0436-0.0361-0.0388
2260.04670.02380.0332
2270.03740.07710.1539
2280.06080.0041-0.0064
2290.0090-0.0511-0.0511
230-0.0433-0.0285-0.0475
2310.04250.07550.0655
2320.04110.07710.0372
2330.07010.03530.0365
2340.06190.0994-0.0010
2350.0307-0.0023-0.0032
236-0.0602-0.0635-0.0401
2370.02030.02390.0794
238-0.0532-0.0546-0.0512
2390.02420.08150.0133
2400.00910.03450.0101
241-0.01590.0242-0.0630
242-0.1097-0.0950-0.0687
243-0.0183-0.0235-0.0303
244-0.0108-0.0220-0.0496
245-0.0582-0.0459-0.0443
2460.09850.13200.2145
2470.06750.08960.1510
2480.00190.0027-0.0417
249-0.02420.0025-0.0594
Source: Emolia / Emilia source archive · SHA-256 82c7725e3e3a3683…
e09KoreanEmolia / Emilia

Emolia KO · KO_B00000_S00030_W000053

Natural speech · Original speech · 8.9 s

다행히도 방송상 렉은 없고 뭐 화면이 급변하는 게임은 아니니까. 그 다음에 이게 아무도 없는 행성이고 플레이어 4가

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Concentration
z +1.49 / +1.72raw 1.74 / 1.91
Contemplation
z +1.05 / +1.32raw 0.71 / 0.87
Contentment
z +1.12 / +0.98raw 0.58 / 0.51
Interest
z +0.48 / +0.53raw 2.02 / 2.07
Confusion
z +0.38 / +0.14raw 0.21 / 0.08
Doubt
z +0.36 / +0.09raw 0.17 / 0.05
Hope Enthusiasm Optimism
z +0.17 / +0.21raw 0.24 / 0.28

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Teacher or didactic style (S_TECH)
z +1.76 / +2.09raw 2.80 / 3.21
Authoritative style (S_AUTH)
z +1.67 / +1.84raw 2.83 / 3.01
Mixed resonance (R_MIXD)
z +1.36 / +1.69raw 2.58 / 2.85
Throat resonance (R_THRT)
z +1.45 / +1.46raw 3.23 / 3.24
Formal style (S_FORM)
z +1.07 / +1.28raw 2.29 / 2.51
Oral resonance (R_ORAL)
z +1.22 / +1.28raw 3.09 / 3.13
Head resonance (R_HEAD)
z +0.94 / +1.25raw 2.92 / 3.17

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z +0.60 / +0.63raw 3.31 / 3.35
Vocal-burst blend (0–10)z -1.38 / -1.34raw 2.91 / 3.01
Content enjoyment (CE)z +0.53 / +0.37raw 5.34 / 5.18
Production quality (PQ)z -0.38 / -0.42raw 6.71 / 6.67
Speech signal quality, calibratedz +0.08 / -0.01raw 3.53 / 3.48
Background quality, calibratedz +0.56 / +0.65raw 3.90 / 3.99
Overall quality, calibratedz +0.54 / +0.55raw 3.17 / 3.18
P.808 MOS estimatez +0.66 / +0.64raw 3.75 / 3.74
score content enjoymentz +0.18 / +0.20raw 4.56 / 4.57
score overall qualityz +0.64 / +0.63raw 2.97 / 2.97
Background Noisez +0.16 / +0.16raw 0.85 / 0.85
Tempo (TEMP)z +0.59 / +0.70raw 2.82 / 2.95
Arousal (AROU)z -0.05 / -0.04raw 2.31 / 2.31
Valence (VALN)z +0.50 / +0.71raw 2.89 / 3.12
Recording quality (RCQL)z -0.04 / -0.03raw 2.72 / 2.73
Background noise (BKGN)z +0.25 / +0.22raw 2.90 / 2.87
talking speedz +0.48 / +0.65raw 10.82 / 11.46
Characters per second (CPS): 7.18 raw · z -0.49

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8380.880
Identity · 250D0.6910.768

Vocal-burst regions

Tiny
Base
0 s8.9 s

Verified burst intervals are unavailable for this clip.

Tiny: 0 regions, count-head estimate 0.02; Base: 0 regions, count-head estimate 0.00.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.007-0.000.004-0.010.0070.570
1Amusementemo_Amusement0.041+0.050.020+0.02-0.0010.885
2Angeremo_Anger0.018+0.020.082+0.130.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.029+0.040.047+0.070.0050.572
4Aweemo_Awe-0.020-0.05-0.011-0.020.0000.427
5Bitternessemo_Bitterness0.022+0.020.020+0.020.0130.415
6Concentrationemo_Concentration1.738+1.491.909+1.720.6340.741
7Confusionemo_Confusion0.207+0.380.075+0.140.0010.540
8Contemplationemo_Contemplation0.708+1.050.869+1.320.0820.598
9Contemptemo_Contempt0.017+0.040.004+0.01-0.0000.464
10Contentmentemo_Contentment0.578+1.120.506+0.980.0000.514
11Disappointmentemo_Disappointment0.008+0.010.040+0.060.0000.700
12Disgustemo_Disgust0.010+0.01-0.003-0.020.0070.489
13Distressemo_Distress0.031+0.040.053+0.070.0000.821
14Doubtemo_Doubt0.174+0.360.046+0.090.0060.466
15Elationemo_Elation0.011-0.000.014+0.000.0120.728
16Embarrassmentemo_Embarrassment-0.017-0.14-0.013-0.130.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.030+0.030.113+0.120.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.047+0.040.035+0.010.0280.539
19Fearemo_Fear-0.034-0.080.009-0.010.0150.610
20Helplessnessemo_Helplessness0.016+0.030.013+0.02-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.243+0.170.279+0.210.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability-0.071-0.31-0.055-0.290.1490.705
23Infatuationemo_Infatuation-0.043-0.05-0.020-0.01-0.0150.533
24Interestemo_Interest2.020+0.482.074+0.531.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness-0.022+0.01-0.043-0.03-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.044-0.030.033-0.060.0590.442
27Longingemo_Longing-0.019-0.170.038-0.080.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.053-0.000.045-0.020.0550.534
29Painemo_Pain0.033+0.060.006+0.01-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy-0.003-0.030.025+0.010.0170.614
31Prideemo_Pride0.044+0.080.100+0.190.0020.509
32Reliefemo_Relief0.040+0.010.010-0.030.0330.681
33Sadnessemo_Sadness0.023+0.030.005+0.010.0000.759
34Sexual Lustemo_Sexual_Lust-0.015-0.09-0.006-0.070.0300.494
35Shameemo_Shame0.010-0.01-0.011-0.050.0130.510
36Sournessemo_Sourness-0.025-0.040.008+0.03-0.0050.439
37Teasingemo_Teasing0.020+0.040.047+0.090.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.102+0.090.031-0.010.0410.697
39Triumphemo_Triumph-0.027-0.04-0.010-0.01-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg2.403-0.472.347-0.512.9841.238
41Arousal (AROU)vn_AROU_reg2.306-0.052.311-0.042.3591.147
42Arousal shift (ARSH)vn_ARSH_reg2.456-0.142.627+0.052.5800.867
43Vocal attack (ATCK)vn_ATCK_reg1.746-0.321.887-0.212.1621.295
44Background noise (BKGN)vn_BKGN_reg2.896+0.252.869+0.222.7030.761
45Brightness (BRGT)vn_BRGT_reg1.758-0.292.067+0.091.9910.813
46Speech chunking (CHNK)vn_CHNK_reg2.880+0.832.934+0.881.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg1.855-0.171.824-0.212.0180.945
48Cognitive load (COGL)vn_COGL_reg2.627+0.502.557+0.442.0941.061
49Dynamic arc (DARC)vn_DARC_reg1.677+0.051.753+0.101.6151.380
50Disfluency (DFLU)vn_DFLU_reg3.009+0.443.392+0.762.4881.184
51Emphasis (EMPH)vn_EMPH_reg3.031+0.283.123+0.342.6071.524
52Esthetics (ESTH)vn_ESTH_reg2.488+0.112.685+0.362.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.137+0.160.122+0.110.0860.322
54Focus (FOCS)vn_FOCS_reg1.600-0.681.582-0.692.3441.102
55Fullness (FULL)vn_FULL_reg2.596+0.922.634+0.961.7780.889
56Perceived gender (GEND)vn_GEND_reg3.695-0.233.894-0.104.0551.588
57Harmonicity (HARM)vn_HARM_reg2.844+0.433.077+0.712.4800.836
58Metallic character (METL)vn_METL_reg2.324+0.732.357+0.771.6210.958
59Pitch range (RANG)vn_RANG_reg2.644-0.302.498-0.462.9140.897
60Recording quality (RCQL)vn_RCQL_reg2.719-0.042.727-0.032.7540.888
61Vocal register (REGS)vn_REGS_reg0.853-0.270.976-0.181.1981.268
62Respiration (RESP)vn_RESP_reg2.266-0.392.095-0.552.6801.056
63Roughness (ROUG)vn_ROUG_reg2.225+0.222.328+0.332.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg2.678+0.642.768+0.752.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg2.922+0.943.173+1.252.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg2.163+0.252.570+0.661.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg2.579+1.362.854+1.691.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg3.054+0.253.159+0.352.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg3.085+1.223.132+1.282.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg3.232+1.453.244+1.462.0700.803
71Smoothness (SMTH)vn_SMTH_reg2.879+0.672.844+0.642.1431.093
72Stance (STNC)vn_STNC_reg2.668+0.032.774+0.122.6331.137
73Speech structure (STRU)vn_STRU_reg2.900+0.802.984+0.882.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg1.945+0.411.848+0.341.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg2.827+1.673.005+1.841.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg1.070-0.350.733-0.591.5601.401
77Casual style (S_CASU)vn_S_CASU_reg3.623-0.133.409-0.303.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg2.944+0.272.890+0.222.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg1.373-1.171.159-1.362.7211.148
80Formal style (S_FORM)vn_S_FORM_reg2.285+1.072.506+1.281.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg3.804+0.763.978+0.902.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg1.803+0.341.400+0.061.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.178-0.410.029-0.670.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg2.489+0.102.596+0.192.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg-0.918-1.65-1.197-1.871.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg1.748-0.541.531-0.702.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg2.803+1.763.212+2.090.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg2.551+0.592.395+0.491.5971.618
89Tempo (TEMP)vn_TEMP_reg2.823+0.592.954+0.702.1171.190
90Tension (TENS)vn_TENS_reg1.572+0.021.458-0.121.5590.852
91Valence (VALN)vn_VALN_reg2.888+0.503.117+0.712.3281.118
92Valence shift (VALS)vn_VALS_reg3.283+0.713.588+0.992.5181.079
93Velocity flux (VFLX)vn_VFLX_reg2.183-0.062.322+0.052.2561.208
94Volatility (VOLT)vn_VOLT_reg1.836-0.851.680-1.022.5860.886
95Vulnerability (VULN)vn_VULN_reg2.059-0.601.974-0.692.6480.984
96Warmth (WARM)vn_WARM_reg2.490-0.102.421-0.172.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_63.312+0.603.349+0.632.3321.621
98Vocal-burst blend (0–10)blend_0_102.907-1.383.013-1.346.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.862-0.690.874-0.651.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.301+0.173.241-0.033.2500.293
101Arousaleiv_extra:Arousal0.961-0.751.063-0.521.2890.435
102Authenticityeiv_extra:Authenticity2.702+0.232.647-0.132.6670.153
103Background Noiseeiv_extra:Background_Noise0.851+0.160.851+0.160.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.092+0.171.162+0.401.0400.303
105Gendereiv_extra:Gender-1.079-0.46-1.017-0.40-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.971+0.571.896+0.271.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive1.692-0.361.739-0.261.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.192+0.142.171+0.082.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous1.136+0.611.027+0.290.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.253-0.090.242-0.120.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.268-0.330.258-0.360.3760.324
112Valenceeiv_extra:Valence-0.054-0.16-0.060-0.170.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached2.099+0.882.096+0.871.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold0.138+0.290.133+0.270.0510.300
115score background qualityeiv_extra:score_background_quality3.613+0.453.681+0.553.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.563+0.184.568+0.204.5070.303
117score overall qualityeiv_extra:score_overall_quality2.975+0.642.973+0.632.6520.507
118score speech qualityeiv_extra:score_speech_quality1.883+0.521.882+0.511.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.340+0.535.179+0.374.7991.018
120Content usefulness (CU)audiobox:CU5.798+0.155.717+0.075.6491.001
121Production complexity (PC)audiobox:PC1.795-0.331.694-0.392.3311.617
122Production quality (PQ)audiobox:PQ6.709-0.386.674-0.427.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw3.981+0.123.901+0.003.9000.669
124Background quality, rawdnsmos:BAK_raw4.026+0.724.162+0.853.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.576+0.613.608+0.643.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.528+0.083.483-0.013.4880.501
127Background quality, calibrateddnsmos:BAK3.905+0.563.992+0.653.3890.928
128Overall quality, calibrateddnsmos:OVRL3.167+0.543.176+0.552.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.750+0.663.738+0.643.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.018+0.030.002+0.000.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.966+0.030.926-0.040.9470.554
132Agevoiceclap_attribute:Age3.032-0.363.148-0.153.2290.547
133Amusementvoiceclap_attribute:Amusement0.142-0.440.119-0.480.3920.572
134Angervoiceclap_attribute:Anger0.323-1.320.264-1.440.9920.506
135Arousalvoiceclap_attribute:Arousal0.832-0.940.968-0.731.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.191-0.660.297-0.440.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.681-0.542.732-0.232.7700.167
138Awevoiceclap_attribute:Awe0.345-0.750.397-0.550.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.534-0.130.578+0.040.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.130-1.960.124-1.990.5920.236
141Concentrationvoiceclap_attribute:Concentration1.720+0.951.861+1.221.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.373+0.651.414+0.741.0690.468
143Confusionvoiceclap_attribute:Confusion0.854+0.630.850+0.620.6430.334
144Contemplationvoiceclap_attribute:Contemplation1.103+0.801.049+0.660.7800.406
145Contemptvoiceclap_attribute:Contempt0.014-2.050.040-1.970.7090.339
146Contentmentvoiceclap_attribute:Contentment0.593+0.020.510-0.250.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.409-1.210.411-1.200.7600.291
148Disgustvoiceclap_attribute:Disgust0.254-1.790.238-1.870.6250.207
149Distressvoiceclap_attribute:Distress0.511-1.190.382-1.560.9270.350
150Doubtvoiceclap_attribute:Doubt0.467-0.360.448-0.460.5350.188
151Elationvoiceclap_attribute:Elation0.296-0.610.399-0.350.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.235-0.230.273-0.110.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.680+0.060.608-0.210.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.397-0.250.272-0.730.4620.261
155Fearvoiceclap_attribute:Fear0.455-0.770.272-1.630.6180.212
156Gendervoiceclap_attribute:Gender-0.830-0.68-0.794-0.64-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.638-0.070.579-0.300.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.936+0.251.984+0.381.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.456-0.630.582-0.310.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability0.235-1.510.206-1.571.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.483-0.380.421-0.560.6080.332
162Interestvoiceclap_attribute:Interest1.733+0.311.832+0.531.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.225-0.600.075-0.920.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.242-1.150.162-1.470.5230.245
165Longingvoiceclap_attribute:Longing0.498-0.180.416-0.560.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.230-1.740.165-1.930.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive1.336-0.741.411-0.611.7750.594
168Painvoiceclap_attribute:Pain0.178-0.930.109-1.250.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.199-1.010.262-0.800.5030.301
170Pridevoiceclap_attribute:Pride0.517-0.750.438-0.980.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.232-0.252.289-0.082.3130.328
172Reliefvoiceclap_attribute:Relief0.849-0.420.615-0.991.0250.415
173Sadnessvoiceclap_attribute:Sadness0.484-0.860.440-1.000.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.777-0.050.750-0.090.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.239-0.950.154-1.150.6350.419
176Shamevoiceclap_attribute:Shame0.468-0.210.371-0.550.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.322+0.450.306+0.410.1600.360
178Sournessvoiceclap_attribute:Sourness0.210-1.700.167-1.850.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.191-1.210.180-1.240.6230.359
180Teasingvoiceclap_attribute:Teasing0.078-1.220.113-1.130.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.433+0.091.069-0.591.3840.529
182Triumphvoiceclap_attribute:Triumph0.423-0.820.403-0.870.7160.359
183Valencevoiceclap_attribute:Valence0.219+0.360.227+0.37-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached2.374+1.322.400+1.371.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold0.169+0.590.183+0.62-0.0660.400
186durationvoiceclap_attribute:duration8.889-0.329.780-0.1810.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.525+0.563.583+0.683.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.636-0.134.641-0.104.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.865+0.562.886+0.612.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.958+0.971.964+1.061.8910.069
191talking speedvoiceclap_attribute:talking_speed10.823+0.4811.460+0.659.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.05150.03810.0730
10.14860.07740.0115
20.00690.00430.0919
3-0.0562-0.0598-0.0908
40.17900.14260.1426
5-0.0193-0.02720.0336
60.01970.04850.0852
7-0.00300.0065-0.0468
8-0.1371-0.1288-0.0861
9-0.0525-0.00560.0284
100.0128-0.02140.0083
11-0.0690-0.1096-0.0764
12-0.0661-0.0866-0.0688
13-0.0751-0.0881-0.0660
14-0.0060-0.0383-0.0746
150.01010.0171-0.0457
16-0.0511-0.04870.0550
17-0.0575-0.04270.0221
180.00200.0242-0.0048
19-0.0599-0.0674-0.0847
20-0.0825-0.0536-0.0897
21-0.0517-0.0475-0.0824
22-0.0800-0.0927-0.0842
23-0.0993-0.0970-0.1022
24-0.0661-0.0674-0.0723
25-0.1002-0.0878-0.0787
26-0.00280.02740.0533
27-0.0821-0.0851-0.0836
28-0.1207-0.1242-0.0987
290.10510.11420.1418
300.26760.33870.3608
31-0.0201-0.0194-0.0425
320.28080.28360.2764
33-0.0571-0.07280.0015
34-0.0948-0.0521-0.0645
350.12150.10730.0131
36-0.00920.0257-0.0068
370.01460.0151-0.0474
38-0.0681-0.0824-0.0894
39-0.0415-0.0310-0.0362
40-0.0739-0.0778-0.0864
41-0.03280.00060.0471
42-0.02770.02250.0098
43-0.0878-0.1050-0.0922
44-0.01920.0219-0.0052
450.02380.04980.1088
460.05750.06480.1207
47-0.0542-0.0805-0.0906
480.0117-0.0114-0.0031
49-0.0850-0.1081-0.0905
50-0.0013-0.0477-0.0626
51-0.00270.0143-0.0178
52-0.0866-0.0851-0.0876
53-0.0587-0.0410-0.0710
54-0.0361-0.03030.0063
55-0.1067-0.1042-0.0862
560.00260.0322-0.0270
57-0.0862-0.0540-0.0983
58-0.1190-0.0889-0.0953
59-0.0507-0.0548-0.0898
600.10750.0908-0.0220
610.02850.03100.0173
620.00500.0041-0.0410
630.01990.01640.0271
640.04620.01400.1153
65-0.0503-0.04430.0589
660.14370.10270.0892
670.0269-0.00420.0278
680.12720.11800.1803
69-0.0591-0.0529-0.0534
70-0.0530-0.0506-0.0831
71-0.0110-0.0024-0.0246
72-0.0223-0.0431-0.0132
730.04130.02220.0150
74-0.1305-0.1058-0.0979
750.11740.04690.1085
760.19130.17470.1331
77-0.0969-0.1127-0.0818
780.06980.15250.1422
79-0.0866-0.0774-0.0777
800.14530.09620.1231
81-0.0302-0.0174-0.0134
82-0.01930.0253-0.0242
830.08870.07360.0186
84-0.1034-0.1050-0.0782
85-0.1014-0.0828-0.0728
860.0086-0.0264-0.0747
87-0.1084-0.0989-0.0243
880.11490.10960.0819
890.08370.07090.0508
900.09650.09660.1566
910.13050.16320.1477
92-0.0940-0.0947-0.0870
93-0.0841-0.0786-0.0875
940.01830.09470.1523
950.06490.08580.0550
96-0.1305-0.1081-0.0837
970.15110.14560.1007
98-0.1067-0.1088-0.0905
990.19290.19470.1683
1000.15760.14410.1470
101-0.0072-0.05100.0157
102-0.0685-0.0671-0.0959
103-0.1051-0.1058-0.0691
1040.02830.06280.1653
105-0.0792-0.0690-0.0864
106-0.0821-0.1165-0.0572
1070.0003-0.0185-0.0303
1080.08830.08120.1041
109-0.0681-0.0575-0.0879
110-0.01810.01760.0425
111-0.1215-0.1226-0.1039
112-0.0384-0.03680.0184
1130.00870.04560.0684
114-0.0685-0.0295-0.0093
1150.04270.03950.0770
116-0.0101-0.0245-0.0064
117-0.0854-0.0858-0.0614
118-0.0562-0.0659-0.0121
119-0.0260-0.0422-0.0488
1200.23480.21460.0720
121-0.0525-0.0410-0.0622
122-0.0956-0.0927-0.0001
1230.0093-0.0705-0.0097
1240.01740.02820.0057
125-0.0837-0.0939-0.0893
126-0.0616-0.0870-0.0772
1270.08870.08580.0975
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.1053-0.0909-0.0552
10.04510.05630.0750
2-0.0916-0.0343-0.0617
30.06020.06770.0437
40.0087-0.0192-0.0457
5-0.0784-0.0727-0.0525
60.08530.0540-0.0344
70.00800.07320.0380
80.02600.0374-0.0577
9-0.0342-0.0100-0.0014
10-0.0529-0.0146-0.0527
11-0.0739-0.0803-0.0679
12-0.0154-0.0207-0.0549
130.02470.0176-0.0443
140.02640.05030.0449
150.11350.00730.0317
160.12770.08330.1858
17-0.0465-0.0525-0.0603
180.04350.00920.0377
19-0.03940.0029-0.0797
20-0.01760.0074-0.0503
21-0.0269-0.0213-0.0465
220.01690.0117-0.0272
23-0.00930.0088-0.0345
24-0.0147-0.0366-0.0428
25-0.06840.00160.0413
260.07930.1030-0.0245
27-0.0638-0.0677-0.0408
280.01880.04950.1152
29-0.0251-0.0520-0.0483
30-0.0050-0.0268-0.0436
31-0.1222-0.0894-0.0630
320.13590.06770.0987
33-0.0780-0.0849-0.0450
340.06700.08080.0411
35-0.0492-0.0285-0.0646
36-0.0802-0.0523-0.0530
37-0.0374-0.0641-0.0347
38-0.1277-0.0869-0.0696
39-0.0206-0.00060.0369
40-0.0518-0.0692-0.0581
41-0.0321-0.0641-0.0576
42-0.0264-0.0069-0.0499
43-0.0074-0.01040.0080
44-0.0463-0.0245-0.0498
45-0.0524-0.0475-0.0576
460.02780.04670.0707
470.0467-0.0111-0.0550
48-0.0907-0.0919-0.0536
49-0.0843-0.0849-0.0894
50-0.0169-0.03710.0009
51-0.0948-0.0798-0.0509
52-0.0716-0.0520-0.0073
53-0.0264-0.0439-0.0486
540.02580.08740.0462
550.17690.15360.1379
560.01410.01400.0238
570.24070.22120.1527
58-0.0524-0.0477-0.0482
590.01300.03560.0632
600.07520.03080.0044
610.07070.11820.1632
62-0.1053-0.0879-0.0441
63-0.0435-0.0503-0.0545
640.03170.0072-0.0303
65-0.0812-0.0970-0.0612
66-0.01520.0302-0.0632
670.03100.0121-0.0528
680.0002-0.0546-0.0300
690.04990.0216-0.0175
70-0.0422-0.0366-0.0493
71-0.0177-0.0251-0.0526
72-0.0310-0.0263-0.0111
730.07660.08330.1483
74-0.0880-0.0818-0.0542
75-0.0317-0.0069-0.0440
760.06380.12730.0759
77-0.0333-0.04650.0267
780.10530.14450.1149
79-0.0825-0.0798-0.0537
80-0.0290-0.05480.0123
810.00210.00290.0242
82-0.0926-0.0934-0.0482
83-0.0798-0.0722-0.0446
84-0.00370.01340.0148
85-0.0182-0.0019-0.0536
860.09260.06920.0108
87-0.0172-0.01170.0175
88-0.04670.03760.0167
890.05240.0304-0.0485
900.07200.05780.0609
910.01450.0119-0.0110
92-0.0716-0.0899-0.0463
93-0.00480.01930.0338
94-0.0054-0.0129-0.0615
95-0.0117-0.0359-0.0109
96-0.0401-0.0586-0.0479
97-0.0397-0.0508-0.0521
98-0.0137-0.0151-0.0566
99-0.0319-0.0624-0.0617
1000.06110.07580.0084
101-0.0394-0.0402-0.0507
102-0.0693-0.0672-0.0532
1030.09390.07530.0084
104-0.0643-0.0975-0.0698
105-0.0199-0.01500.0201
106-0.0458-0.0314-0.0413
107-0.0119-0.0609-0.0480
1080.08710.0641-0.0279
1090.06570.05150.0554
110-0.0556-0.0945-0.0638
1110.10400.06870.0106
112-0.0362-0.0298-0.0433
113-0.0739-0.0525-0.0914
1140.02380.03860.1129
115-0.0527-0.0412-0.0380
1160.05790.08130.1392
1170.02230.00820.0551
1180.04860.05990.0795
119-0.05020.00040.0071
120-0.01060.04500.0538
1210.01350.0326-0.0084
122-0.0277-0.0568-0.0506
123-0.0376-0.0333-0.0124
124-0.0588-0.0707-0.0534
125-0.0198-0.0092-0.0521
1260.07750.03860.0800
127-0.0388-0.0336-0.0551
128-0.0263-0.0274-0.0101
129-0.0771-0.0634-0.0395
130-0.0821-0.0940-0.0538
131-0.0547-0.0712-0.0469
1320.02290.04600.0493
1330.01080.0192-0.0530
1340.06250.05150.1266
135-0.0752-0.0692-0.0539
1360.0012-0.0343-0.0574
137-0.0894-0.0838-0.0498
1380.0088-0.0356-0.0473
139-0.0312-0.0149-0.0424
140-0.0784-0.03180.0039
1410.11440.08080.0177
142-0.0419-0.0657-0.0463
143-0.0026-0.0117-0.0220
144-0.0561-0.0551-0.0558
145-0.0666-0.0639-0.0681
1460.0050-0.0058-0.0066
147-0.01090.0049-0.0012
148-0.0035-0.0235-0.0591
149-0.0866-0.0652-0.0690
150-0.0220-0.0139-0.0527
1510.0117-0.0123-0.0611
1520.06110.0168-0.0561
153-0.00050.0563-0.0649
154-0.01980.0087-0.0522
155-0.0035-0.0427-0.0699
156-0.0319-0.0356-0.0537
1570.15590.18890.1256
158-0.0378-0.0447-0.0630
159-0.0454-0.0304-0.0627
160-0.1044-0.0950-0.0509
161-0.03330.0071-0.0529
162-0.0584-0.0904-0.0622
1630.03530.06670.0668
1640.00600.0151-0.0466
1650.03510.08890.1335
166-0.0022-0.0302-0.0623
1670.10810.07270.1727
168-0.0926-0.0356-0.0757
169-0.0179-0.0662-0.0611
170-0.0730-0.0753-0.0502
1710.08620.04420.0492
172-0.1094-0.0798-0.0621
1730.13500.09900.0588
174-0.0283-0.0490-0.0333
175-0.0848-0.0515-0.0887
1760.06750.0354-0.0456
177-0.0317-0.0389-0.0442
1780.08660.10660.0722
1790.09350.12020.0336
180-0.0527-0.0596-0.0608
181-0.0342-0.00930.0073
182-0.0419-0.0379-0.0463
1830.00130.01790.0206
1840.0027-0.0364-0.0344
1850.05810.04170.0932
1860.0046-0.0261-0.0646
1870.04290.05280.0322
1880.04220.05130.0214
1890.00520.07780.1396
1900.0515-0.0100-0.0311
1910.00620.03660.0226
192-0.0807-0.04800.0069
193-0.0358-0.0798-0.0522
1940.03880.0596-0.0521
1950.00720.01380.0097
1960.03030.0626-0.0045
197-0.0426-0.0157-0.0750
198-0.0766-0.0773-0.0518
199-0.0518-0.04140.0381
200-0.1126-0.0753-0.0619
2010.0693-0.0095-0.0505
202-0.0752-0.0414-0.0748
2030.0017-0.0333-0.0321
204-0.0513-0.0470-0.0627
205-0.0889-0.0470-0.0894
206-0.0515-0.0174-0.0552
207-0.02960.02840.1013
208-0.0761-0.0798-0.0556
2090.00290.0188-0.0187
210-0.0217-0.0576-0.0697
211-0.0472-0.0331-0.0412
2120.04560.0399-0.0212
2130.17510.14350.1445
2140.08020.14650.0745
2150.17600.17070.1527
216-0.0280-0.0149-0.0641
2170.08070.07930.1520
218-0.0698-0.0604-0.0280
219-0.0670-0.0904-0.0698
220-0.0780-0.1061-0.0591
221-0.01470.0005-0.0465
2220.00550.0098-0.0388
2230.0015-0.0384-0.0430
224-0.0135-0.0294-0.0489
2250.05590.0265-0.0446
2260.09620.08890.0652
2270.02850.05830.0837
228-0.0894-0.0697-0.0867
2290.0219-0.0045-0.0009
2300.01410.05860.1074
2310.10300.13840.0695
2320.05840.15360.1298
233-0.1231-0.1576-0.1094
2340.04950.04390.0410
235-0.0693-0.04020.0067
236-0.0652-0.0743-0.0460
2370.01040.0050-0.0547
2380.05770.0326-0.0588
2390.03420.00940.0058
240-0.0688-0.0748-0.0546
2410.0024-0.02840.0079
2420.0100-0.0407-0.0789
243-0.1195-0.1293-0.0688
2440.03420.0429-0.0569
245-0.0137-0.0414-0.0509
246-0.0406-0.0778-0.0434
247-0.0305-0.0178-0.0389
248-0.0522-0.0823-0.0355
2490.02990.01580.1168
Source: Emolia / Emilia source archive · SHA-256 bc068846a09df6c2…
e10KoreanEmolia / Emilia

Emolia KO · KO_B00000_S00030_W000205

Natural speech · Original speech · 7.1 s

그 마찰력 없는 움직임, 허공을 부유하고 있는 움직임을 들어줘, 예. WSD의 움직임이.

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Contemplation
z +1.48 / +0.93raw 0.97 / 0.64
Concentration
z +1.43 / +1.24raw 1.69 / 1.56
Doubt
z +1.03 / +0.71raw 0.48 / 0.33
Confusion
z +0.81 / +0.68raw 0.44 / 0.37
Anger
z +0.18 / +0.39raw 0.12 / 0.24
Contentment
z +0.35 / +0.14raw 0.18 / 0.07
Interest
z +0.25 / +0.17raw 1.82 / 1.74

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Authoritative style (S_AUTH)
z +1.85 / +1.92raw 3.01 / 3.09
Mixed resonance (R_MIXD)
z +1.73 / +1.87raw 2.89 / 3.01
Teacher or didactic style (S_TECH)
z +1.61 / +1.62raw 2.62 / 2.63
Formal style (S_FORM)
z +1.23 / +1.35raw 2.45 / 2.58
Oral resonance (R_ORAL)
z +1.32 / +1.34raw 3.16 / 3.18
Throat resonance (R_THRT)
z +1.17 / +0.98raw 3.01 / 2.86
Head resonance (R_HEAD)
z +0.66 / +0.93raw 2.69 / 2.91

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z +0.78 / +0.93raw 3.59 / 3.84
Vocal-burst blend (0–10)z -0.89 / -0.63raw 4.24 / 4.95
Content enjoyment (CE)z +0.34 / +0.22raw 5.14 / 5.02
Production quality (PQ)z -0.21 / -0.14raw 6.86 / 6.92
Speech signal quality, calibratedz -0.10 / -0.11raw 3.44 / 3.43
Background quality, calibratedz +0.63 / +0.67raw 3.97 / 4.01
Overall quality, calibratedz +0.48 / +0.50raw 3.13 / 3.14
P.808 MOS estimatez +0.94 / +0.74raw 3.89 / 3.79
score content enjoymentz -0.21 / -0.28raw 4.44 / 4.42
score overall qualityz +0.67 / +0.67raw 2.99 / 2.99
Background Noisez -0.27 / -0.08raw 0.74 / 0.79
Tempo (TEMP)z +0.65 / +0.77raw 2.89 / 3.03
Arousal (AROU)z -0.06 / -0.12raw 2.29 / 2.22
Valence (VALN)z +0.49 / +0.45raw 2.87 / 2.83
Recording quality (RCQL)z +0.10 / +0.20raw 2.84 / 2.93
Background noise (BKGN)z +0.47 / +0.33raw 3.06 / 2.95
talking speedz +0.67 / +1.12raw 11.53 / 13.22
Characters per second (CPS): 6.89 raw · z -0.57

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8270.856
Identity · 250D0.8030.811

Vocal-burst regions

Tiny
Base
0 s7.1 s

Verified burst intervals are unavailable for this clip.

Tiny: 0 regions, count-head estimate 0.08; Base: 0 regions, count-head estimate 0.02.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.031+0.040.041+0.060.0070.570
1Amusementemo_Amusement-0.027-0.03-0.010-0.01-0.0010.885
2Angeremo_Anger0.116+0.180.243+0.390.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.020+0.030.018+0.020.0050.572
4Aweemo_Awe0.007+0.020.002+0.000.0000.427
5Bitternessemo_Bitterness0.014+0.000.003-0.020.0130.415
6Concentrationemo_Concentration1.692+1.431.555+1.240.6340.741
7Confusionemo_Confusion0.439+0.810.366+0.680.0010.540
8Contemplationemo_Contemplation0.968+1.480.636+0.930.0820.598
9Contemptemo_Contempt0.047+0.100.005+0.01-0.0000.464
10Contentmentemo_Contentment0.180+0.350.071+0.140.0000.514
11Disappointmentemo_Disappointment0.019+0.03-0.012-0.020.0000.700
12Disgustemo_Disgust0.055+0.100.006-0.000.0070.489
13Distressemo_Distress0.028+0.030.022+0.030.0000.821
14Doubtemo_Doubt0.485+1.030.334+0.710.0060.466
15Elationemo_Elation0.017+0.01-0.067-0.110.0120.728
16Embarrassmentemo_Embarrassment0.022-0.050.012-0.070.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.154+0.160.154+0.160.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.022-0.010.027-0.000.0280.539
19Fearemo_Fear0.011-0.010.002-0.020.0150.610
20Helplessnessemo_Helplessness0.030+0.05-0.009-0.01-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.080-0.050.053-0.080.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.079-0.100.248+0.140.1490.705
23Infatuationemo_Infatuation-0.049-0.06-0.006+0.02-0.0150.533
24Interestemo_Interest1.815+0.251.744+0.171.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness-0.015+0.02-0.005+0.04-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.027-0.070.037-0.050.0590.442
27Longingemo_Longing0.144+0.090.058-0.050.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.066+0.020.041-0.030.0550.534
29Painemo_Pain0.017+0.030.021+0.04-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy-0.025-0.070.030+0.020.0170.614
31Prideemo_Pride-0.019-0.04-0.008-0.020.0020.509
32Reliefemo_Relief0.007-0.040.021-0.020.0330.681
33Sadnessemo_Sadness0.029+0.040.017+0.020.0000.759
34Sexual Lustemo_Sexual_Lust-0.008-0.08-0.052-0.170.0300.494
35Shameemo_Shame0.002-0.020.010-0.010.0130.510
36Sournessemo_Sourness0.010+0.03-0.011-0.01-0.0050.439
37Teasingemo_Teasing0.009+0.020.028+0.050.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.045+0.010.042+0.000.0410.697
39Triumphemo_Triumph-0.055-0.09-0.019-0.02-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg2.419-0.462.344-0.522.9841.238
41Arousal (AROU)vn_AROU_reg2.289-0.062.217-0.122.3591.147
42Arousal shift (ARSH)vn_ARSH_reg2.348-0.272.529-0.062.5800.867
43Vocal attack (ATCK)vn_ATCK_reg1.978-0.141.974-0.152.1621.295
44Background noise (BKGN)vn_BKGN_reg3.064+0.472.953+0.332.7030.761
45Brightness (BRGT)vn_BRGT_reg1.943-0.062.092+0.121.9910.813
46Speech chunking (CHNK)vn_CHNK_reg2.827+0.792.890+0.841.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg1.851-0.181.857-0.172.0180.945
48Cognitive load (COGL)vn_COGL_reg2.754+0.622.559+0.442.0941.061
49Dynamic arc (DARC)vn_DARC_reg1.575-0.031.515-0.071.6151.380
50Disfluency (DFLU)vn_DFLU_reg2.868+0.323.175+0.582.4881.184
51Emphasis (EMPH)vn_EMPH_reg2.991+0.252.994+0.252.6071.524
52Esthetics (ESTH)vn_ESTH_reg2.513+0.142.562+0.202.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.051-0.110.022-0.200.0860.322
54Focus (FOCS)vn_FOCS_reg1.786-0.511.881-0.422.3441.102
55Fullness (FULL)vn_FULL_reg2.307+0.592.424+0.731.7780.889
56Perceived gender (GEND)vn_GEND_reg4.167+0.074.273+0.144.0551.588
57Harmonicity (HARM)vn_HARM_reg2.498+0.022.454-0.032.4800.836
58Metallic character (METL)vn_METL_reg2.413+0.832.275+0.681.6210.958
59Pitch range (RANG)vn_RANG_reg2.623-0.322.649-0.292.9140.897
60Recording quality (RCQL)vn_RCQL_reg2.841+0.102.927+0.202.7540.888
61Vocal register (REGS)vn_REGS_reg0.476-0.570.544-0.521.1981.268
62Respiration (RESP)vn_RESP_reg2.040-0.611.961-0.682.6801.056
63Roughness (ROUG)vn_ROUG_reg2.379+0.392.353+0.362.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg2.831+0.822.847+0.842.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg2.687+0.662.909+0.932.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg2.189+0.282.331+0.421.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg2.893+1.733.011+1.871.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg2.791-0.002.805+0.012.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg3.161+1.323.178+1.342.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg3.012+1.172.861+0.982.0700.803
71Smoothness (SMTH)vn_SMTH_reg2.727+0.532.744+0.552.1431.093
72Stance (STNC)vn_STNC_reg2.690+0.052.699+0.062.6331.137
73Speech structure (STRU)vn_STRU_reg2.969+0.872.823+0.722.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg2.185+0.581.856+0.341.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg3.013+1.853.094+1.921.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg0.899-0.470.925-0.451.5601.401
77Casual style (S_CASU)vn_S_CASU_reg3.524-0.213.481-0.243.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg3.380+0.603.400+0.612.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg1.680-0.911.747-0.852.7211.148
80Formal style (S_FORM)vn_S_FORM_reg2.451+1.232.577+1.351.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg3.265+0.343.412+0.462.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg1.737+0.301.623+0.221.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.467+0.100.377-0.050.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg2.434+0.062.606+0.192.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg-0.141-1.05-0.091-1.011.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg2.174-0.222.258-0.162.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg2.622+1.612.632+1.620.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg2.238+0.402.137+0.331.5971.618
89Tempo (TEMP)vn_TEMP_reg2.894+0.653.034+0.772.1171.190
90Tension (TENS)vn_TENS_reg1.496-0.071.377-0.211.5590.852
91Valence (VALN)vn_VALN_reg2.870+0.492.835+0.452.3281.118
92Valence shift (VALS)vn_VALS_reg3.304+0.733.210+0.642.5181.079
93Velocity flux (VFLX)vn_VFLX_reg2.290+0.032.451+0.162.2561.208
94Volatility (VOLT)vn_VOLT_reg1.938-0.731.908-0.772.5860.886
95Vulnerability (VULN)vn_VULN_reg1.734-0.931.881-0.782.6480.984
96Warmth (WARM)vn_WARM_reg2.600+0.022.637+0.062.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_63.594+0.783.836+0.932.3321.621
98Vocal-burst blend (0–10)blend_0_104.245-0.894.945-0.636.6762.738
99Legacy R_quality (no P3 test labels)R_quality1.036-0.171.143+0.141.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.331+0.273.301+0.173.2500.293
101Arousaleiv_extra:Arousal0.983-0.701.080-0.481.2890.435
102Authenticityeiv_extra:Authenticity2.664-0.022.599-0.442.6670.153
103Background Noiseeiv_extra:Background_Noise0.737-0.270.787-0.080.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.378+1.121.389+1.161.0400.303
105Gendereiv_extra:Gender-0.904-0.28-0.911-0.29-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.877+0.201.792-0.141.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive1.661-0.421.647-0.451.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.071-0.221.989-0.472.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous1.129+0.591.148+0.640.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.184-0.300.090-0.600.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.390+0.040.261-0.350.3760.324
112Valenceeiv_extra:Valence0.013+0.020.019+0.040.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.943+0.521.970+0.581.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold0.064+0.040.129+0.260.0510.300
115score background qualityeiv_extra:score_background_quality3.645+0.503.708+0.593.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.445-0.214.421-0.284.5070.303
117score overall qualityeiv_extra:score_overall_quality2.992+0.672.992+0.672.6520.507
118score speech qualityeiv_extra:score_speech_quality1.890+0.561.915+0.731.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.141+0.345.024+0.224.7991.018
120Content usefulness (CU)audiobox:CU5.673+0.025.720+0.075.6491.001
121Production complexity (PC)audiobox:PC1.751-0.361.852-0.302.3311.617
122Production quality (PQ)audiobox:PQ6.861-0.216.923-0.147.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw3.828-0.113.815-0.133.9000.669
124Background quality, rawdnsmos:BAK_raw4.109+0.804.166+0.853.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.519+0.533.545+0.573.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.439-0.103.435-0.113.4880.501
127Background quality, calibrateddnsmos:BAK3.974+0.634.014+0.673.3890.928
128Overall quality, calibrateddnsmos:OVRL3.129+0.483.143+0.502.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.888+0.943.787+0.743.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.078+0.120.018+0.030.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.992+0.080.801-0.260.9470.554
132Agevoiceclap_attribute:Age3.157-0.133.233+0.013.2290.547
133Amusementvoiceclap_attribute:Amusement0.115-0.490.146-0.430.3920.572
134Angervoiceclap_attribute:Anger0.315-1.340.375-1.220.9920.506
135Arousalvoiceclap_attribute:Arousal0.997-0.681.080-0.551.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.307-0.420.451-0.120.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.687-0.502.736-0.212.7700.167
138Awevoiceclap_attribute:Awe0.304-0.910.396-0.550.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.429-0.520.417-0.570.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.186-1.720.227-1.550.5920.236
141Concentrationvoiceclap_attribute:Concentration1.533+0.601.610+0.751.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.507+0.941.511+0.941.0690.468
143Confusionvoiceclap_attribute:Confusion1.017+1.120.989+1.040.6430.334
144Contemplationvoiceclap_attribute:Contemplation1.142+0.891.041+0.640.7800.406
145Contemptvoiceclap_attribute:Contempt0.160-1.620.290-1.240.7090.339
146Contentmentvoiceclap_attribute:Contentment0.467-0.390.287-0.980.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.536-0.770.523-0.810.7600.291
148Disgustvoiceclap_attribute:Disgust0.242-1.850.308-1.530.6250.207
149Distressvoiceclap_attribute:Distress0.582-0.990.533-1.130.9270.350
150Doubtvoiceclap_attribute:Doubt0.637+0.540.640+0.560.5350.188
151Elationvoiceclap_attribute:Elation0.283-0.640.322-0.550.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.257-0.160.311+0.010.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.729+0.250.732+0.260.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.226-0.900.189-1.050.4620.261
155Fearvoiceclap_attribute:Fear0.414-0.960.327-1.370.6180.212
156Gendervoiceclap_attribute:Gender-0.673-0.52-0.704-0.55-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.614-0.160.559-0.370.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.761-0.211.822-0.051.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.513-0.480.462-0.610.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability0.413-1.180.472-1.071.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.433-0.530.398-0.630.6080.332
162Interestvoiceclap_attribute:Interest1.665+0.151.735+0.311.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.175-0.710.089-0.890.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.329-0.790.264-1.060.5230.245
165Longingvoiceclap_attribute:Longing0.520-0.080.472-0.300.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.303-1.530.352-1.390.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive1.350-0.721.425-0.591.7750.594
168Painvoiceclap_attribute:Pain0.158-1.020.145-1.080.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.196-1.020.206-0.990.5030.301
170Pridevoiceclap_attribute:Pride0.500-0.800.389-1.120.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.246-0.202.277-0.112.3130.328
172Reliefvoiceclap_attribute:Relief0.595-1.040.498-1.271.0250.415
173Sadnessvoiceclap_attribute:Sadness0.530-0.710.575-0.560.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.798-0.010.869+0.110.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.157-1.140.196-1.050.6350.419
176Shamevoiceclap_attribute:Shame0.472-0.190.431-0.340.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.297+0.380.258+0.270.1600.360
178Sournessvoiceclap_attribute:Sourness0.342-1.250.326-1.310.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.244-1.060.258-1.020.6230.359
180Teasingvoiceclap_attribute:Teasing0.146-1.040.202-0.890.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.371-0.021.044-0.641.3840.529
182Triumphvoiceclap_attribute:Triumph0.406-0.860.417-0.830.7160.359
183Valencevoiceclap_attribute:Valence0.074+0.130.068+0.12-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached2.248+1.102.213+1.031.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold0.120+0.470.096+0.41-0.0660.400
186durationvoiceclap_attribute:duration7.330-0.567.454-0.5410.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.569+0.653.550+0.613.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.635-0.144.636-0.134.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.832+0.482.828+0.472.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.999+1.561.994+1.481.8910.069
191talking speedvoiceclap_attribute:talking_speed11.534+0.6713.219+1.129.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.04840.07110.0814
10.05200.0638-0.0666
20.04150.01320.0716
3-0.0747-0.0450-0.0939
40.16840.15300.1570
5-0.00640.01300.0795
60.0264-0.00630.0558
7-0.0455-0.0551-0.0561
8-0.1214-0.1653-0.1200
9-0.1077-0.06340.0202
10-0.0292-0.0703-0.0136
11-0.0812-0.1102-0.0932
12-0.1055-0.1131-0.0684
13-0.0652-0.0798-0.0379
140.00930.0209-0.0444
15-0.02980.0096-0.0649
16-0.0561-0.08190.0305
17-0.0618-0.01310.0579
18-0.07320.0158-0.0187
19-0.0683-0.0787-0.0876
20-0.0842-0.0740-0.0927
21-0.0941-0.0721-0.0852
22-0.0415-0.0754-0.0871
23-0.0914-0.1095-0.1096
24-0.0336-0.0276-0.0748
25-0.0736-0.0750-0.0814
26-0.01420.03370.0858
27-0.0789-0.0765-0.0864
28-0.1032-0.1240-0.1021
290.09860.12330.1525
300.19270.29730.3103
31-0.0493-0.0299-0.0441
320.34440.27410.2817
33-0.0929-0.07580.0213
34-0.0596-0.02340.0117
350.01960.06820.0053
360.04270.07110.0346
37-0.0486-0.0281-0.0783
38-0.0907-0.0779-0.0925
39-0.0055-0.0036-0.0334
40-0.0880-0.0696-0.0893
41-0.04860.0071-0.0107
420.04670.02390.0349
43-0.0634-0.0885-0.0953
44-0.0660-0.0254-0.0431
450.05460.06960.1058
460.06220.02430.1065
47-0.0873-0.0924-0.0936
48-0.00260.0076-0.0023
49-0.0283-0.0620-0.0936
50-0.0311-0.0642-0.0329
510.0645-0.0008-0.0374
52-0.0812-0.0754-0.1052
53-0.0793-0.0678-0.0944
54-0.0706-0.02900.0084
55-0.0785-0.0921-0.0891
56-0.0736-0.0591-0.0878
57-0.0971-0.0743-0.1016
58-0.1320-0.1088-0.0985
59-0.0759-0.0779-0.0929
600.12210.1457-0.0080
61-0.0127-0.0381-0.0688
620.09070.0653-0.0277
63-0.0377-0.0022-0.0092
640.04100.02140.1102
65-0.0237-0.03730.0340
660.13580.11090.0900
670.09520.05000.0986
680.08120.07760.1449
69-0.0569-0.0761-0.0848
70-0.0444-0.0053-0.0259
71-0.0448-0.0468-0.0377
720.02300.01800.0092
730.0016-0.0301-0.0394
74-0.1146-0.1037-0.1012
750.11150.10800.1431
760.12900.14720.1425
77-0.0774-0.0866-0.0705
780.09070.12180.1447
79-0.0497-0.0750-0.0803
800.18130.16170.1725
81-0.0252-0.0823-0.0628
82-0.0396-0.0054-0.0236
830.06900.0464-0.0050
84-0.0964-0.1066-0.0974
85-0.1009-0.1022-0.0872
86-0.1032-0.0877-0.0940
87-0.0713-0.1066-0.0098
880.06300.06270.0251
890.01260.04790.0657
900.15860.17620.1401
91-0.0360-0.00790.0895
92-0.0431-0.0772-0.0899
93-0.0971-0.0703-0.0905
940.09330.12250.1349
950.11990.11020.0217
96-0.1138-0.1073-0.0623
970.18890.20740.0920
98-0.0918-0.1182-0.0936
990.15400.12690.1025
1000.19730.14430.1252
101-0.0470-0.06240.0290
102-0.1085-0.0852-0.0991
103-0.0755-0.0895-0.0544
1040.07090.05950.1362
105-0.0835-0.1001-0.0893
106-0.0690-0.0914-0.0247
107-0.0429-0.0319-0.0206
1080.08760.07140.1298
109-0.0804-0.0558-0.0908
1100.04890.04420.0772
111-0.1085-0.1131-0.1074
1120.02170.02320.0398
113-0.05540.01010.0297
114-0.0694-0.06450.0158
1150.03170.04890.0519
116-0.0182-0.0247-0.0794
117-0.0903-0.0885-0.0164
118-0.0994-0.1059-0.0696
119-0.0812-0.0682-0.0844
1200.16620.14720.0792
121-0.0092-0.0761-0.0939
122-0.0899-0.0874-0.0047
1230.02290.01700.0215
124-0.0298-0.0008-0.0406
125-0.0721-0.0544-0.0923
126-0.0804-0.0852-0.0798
1270.08080.09430.0632
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.0931-0.0912-0.0576
10.04100.02210.1057
2-0.0656-0.0485-0.0511
30.06740.06050.0527
4-0.0132-0.0131-0.0374
5-0.0686-0.0794-0.0547
60.06260.0370-0.0312
70.02360.02750.0381
80.00590.0316-0.0769
9-0.0182-0.0456-0.0095
10-0.0230-0.0188-0.0550
11-0.0543-0.0781-0.0709
12-0.0704-0.0539-0.0573
13-0.0093-0.0059-0.0462
14-0.00450.0032-0.0161
150.04500.01830.0766
160.17330.17880.1967
17-0.0729-0.1097-0.0629
180.01740.00230.0453
19-0.01380.0086-0.0495
20-0.01810.0015-0.0524
21-0.0510-0.0309-0.0486
220.0233-0.0221-0.0024
23-0.02620.0296-0.0078
24-0.0639-0.0456-0.0447
25-0.0087-0.01260.0174
260.08110.0831-0.0538
27-0.0500-0.0636-0.0426
280.00830.00620.0534
29-0.0498-0.0627-0.0504
30-0.0281-0.0187-0.0358
31-0.0716-0.0767-0.0658
320.19900.19230.1148
33-0.0300-0.0849-0.0470
340.01590.04290.0022
35-0.0403-0.0332-0.0344
36-0.1000-0.0831-0.0554
37-0.0442-0.0659-0.0362
38-0.0974-0.0727-0.0727
39-0.0239-0.0063-0.0245
40-0.0622-0.0641-0.0607
41-0.0012-0.0325-0.0602
42-0.0315-0.0499-0.0521
430.06310.06050.0327
44-0.0420-0.0404-0.0256
45-0.0776-0.0557-0.0601
460.03620.03300.1099
47-0.0020-0.0090-0.0267
48-0.0789-0.0849-0.0559
49-0.0798-0.0912-0.0932
50-0.0222-0.0467-0.0074
51-0.0678-0.1052-0.0531
52-0.0751-0.05910.0034
53-0.0373-0.0320-0.0507
540.05790.12010.0883
550.11410.13630.0820
56-0.0480-0.0246-0.0714
570.19480.21220.1569
58-0.0147-0.0111-0.0503
59-0.00100.04060.0483
600.04200.0343-0.0349
610.14500.13270.1548
62-0.1201-0.1020-0.0460
63-0.0252-0.0458-0.0569
640.05430.02810.0149
65-0.0918-0.1101-0.0639
66-0.0643-0.0237-0.0660
670.02500.0167-0.0551
68-0.0161-0.0618-0.0229
69-0.0326-0.0388-0.0399
70-0.0401-0.0623-0.0488
71-0.0609-0.0609-0.0549
72-0.0305-0.0386-0.0341
730.10420.10340.1360
74-0.0867-0.0794-0.0565
75-0.0772-0.0517-0.0762
760.07760.10380.0567
770.00510.01380.0700
780.12700.11100.0441
79-0.0875-0.0790-0.0560
800.04050.02890.0604
81-0.00140.00950.0134
82-0.0982-0.1043-0.0503
83-0.0631-0.0664-0.0465
84-0.0245-0.0225-0.0035
85-0.0418-0.0200-0.0559
860.09520.09250.1085
87-0.0144-0.0488-0.0388
880.03430.0300-0.0080
89-0.0609-0.0071-0.0506
900.01280.04380.0216
91-0.0386-0.0077-0.0043
92-0.0772-0.0989-0.0483
930.01750.04180.0587
94-0.0746-0.0641-0.0905
95-0.0725-0.0481-0.0584
96-0.0468-0.0521-0.0500
970.03820.00720.0054
98-0.0145-0.0172-0.0567
99-0.0365-0.0467-0.0443
1000.10720.10380.0427
101-0.0922-0.0591-0.0529
102-0.0057-0.02200.0002
1030.00410.0288-0.0513
104-0.0768-0.0790-0.0729
1050.0047-0.00480.0033
106-0.0125-0.0018-0.0431
107-0.0358-0.0357-0.0501
1080.02810.0359-0.0070
1090.08710.05420.0699
110-0.0480-0.0133-0.0666
1110.00520.0244-0.0052
112-0.0253-0.0393-0.0451
113-0.0824-0.0754-0.0953
114-0.00730.01040.0392
115-0.0408-0.0219-0.0396
1160.07680.09710.0956
117-0.0236-0.02210.0185
1180.02240.06410.0761
119-0.0423-0.0451-0.0115
1200.00870.01240.0583
121-0.0515-0.0318-0.0602
122-0.0643-0.0894-0.0528
1230.0186-0.0056-0.0031
124-0.0609-0.0458-0.0558
125-0.0375-0.0010-0.0544
1260.10810.09750.1181
127-0.0510-0.0600-0.0575
128-0.0472-0.0447-0.0696
129-0.0828-0.0695-0.0593
130-0.0862-0.0858-0.0562
131-0.0541-0.0524-0.0489
1320.01810.05060.0444
133-0.00160.0012-0.0553
1340.0081-0.00130.1179
135-0.0158-0.0533-0.0563
1360.0129-0.0519-0.0599
137-0.1133-0.0822-0.0519
138-0.0332-0.0363-0.0493
139-0.0313-0.0121-0.0509
140-0.0298-0.0418-0.0112
1410.05580.09160.0977
142-0.0440-0.0485-0.0483
143-0.0476-0.0108-0.0260
144-0.0867-0.0740-0.0583
145-0.0652-0.0695-0.0710
1460.0128-0.01520.0240
1470.05130.02690.0826
148-0.0487-0.0038-0.0617
149-0.0918-0.0817-0.0720
150-0.0257-0.0393-0.0550
151-0.0547-0.0311-0.0456
1520.0746-0.0104-0.0141
1530.00130.0184-0.0326
154-0.0579-0.0267-0.0545
155-0.0588-0.0488-0.0729
156-0.0538-0.0316-0.0560
1570.12270.13810.1053
158-0.0001-0.0096-0.0696
159-0.0609-0.0636-0.0655
160-0.0939-0.1065-0.0531
161-0.0324-0.0571-0.0497
162-0.0618-0.0772-0.0649
163-0.0129-0.02890.0228
1640.0082-0.0073-0.0491
1650.08150.06770.0978
166-0.0307-0.0257-0.0651
1670.01320.02880.1406
168-0.0871-0.0772-0.0790
169-0.0704-0.0438-0.0638
170-0.0849-0.0957-0.0523
1710.16470.08030.0548
172-0.0871-0.0925-0.0648
1730.09700.08890.0517
174-0.0105-0.0204-0.0473
175-0.0871-0.0740-0.0762
176-0.0100-0.0090-0.0475
177-0.0367-0.0429-0.0461
1780.09090.03950.0400
1790.04140.0673-0.0170
180-0.0789-0.0952-0.0618
181-0.0423-0.0510-0.0420
182-0.0035-0.0314-0.0483
183-0.01440.02580.0224
184-0.0571-0.0609-0.0568
1850.01400.02750.0914
1860.0091-0.0032-0.0507
187-0.01720.01370.0330
1880.02850.00920.0106
1890.05380.05100.0941
1900.0251-0.0262-0.0169
191-0.0506-0.0485-0.0518
192-0.0751-0.0524-0.0425
1930.11070.02090.0661
1940.08620.0578-0.0053
1950.09090.04900.1082
1960.02340.0225-0.0240
1970.0256-0.0048-0.0126
198-0.0738-0.0731-0.0540
199-0.00130.00180.0193
200-0.0691-0.0740-0.0646
201-0.0571-0.0758-0.0526
202-0.0901-0.0557-0.0867
2030.05920.04400.0478
204-0.0957-0.1165-0.0654
205-0.0948-0.0966-0.0818
206-0.0305-0.0420-0.0576
2070.03430.05710.1451
208-0.0884-0.0898-0.0580
209-0.0362-0.0300-0.0165
210-0.0729-0.0889-0.0728
211-0.0243-0.0488-0.0430
212-0.0418-0.0085-0.0591
2130.14840.16160.1382
2140.11240.10880.0523
2150.11500.14080.1852
216-0.0255-0.0399-0.0669
2170.06350.06640.1442
218-0.0474-0.0591-0.0124
219-0.0678-0.0880-0.0728
220-0.0725-0.0889-0.0617
2210.00030.0492-0.0468
2220.02960.0118-0.0656
2230.0157-0.0492-0.0448
224-0.0123-0.0388-0.0510
2250.06130.0424-0.0465
2260.06220.07180.0567
2270.03580.02750.0878
228-0.0845-0.0790-0.0905
2290.0032-0.00960.0240
2300.03730.02620.0834
2310.06740.11190.0756
2320.06610.07180.0860
233-0.1476-0.1264-0.1141
2340.01950.04970.0323
235-0.0029-0.02070.0563
236-0.0504-0.0627-0.0480
2370.03240.04360.0097
2380.06090.03410.0173
2390.07590.02100.0145
240-0.0708-0.0889-0.0570
2410.0371-0.0264-0.0010
242-0.0117-0.0418-0.0823
243-0.1227-0.1237-0.0718
244-0.02540.0080-0.0411
245-0.0219-0.0497-0.0531
246-0.0513-0.0704-0.0453
2470.00770.00220.0098
248-0.0300-0.03700.0200
2490.00710.00430.0788
Source: Emolia / Emilia source archive · SHA-256 7a85fbdf8f787486…
e11ChineseEmolia / Emilia

Emolia ZH · ZH_B00000_S00110_W000032

Natural speech · Original speech · 8.4 s

万事达卡提出三管齐下的加密支持战略,cryto点com app已启用通过sana网络存取USDT的功能。

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Concentration
z +1.44 / +1.36raw 1.70 / 1.64
Contemplation
z +0.53 / +0.41raw 0.40 / 0.33
Contentment
z +0.27 / +0.44raw 0.14 / 0.23
Emotional Numbness
z +0.15 / +0.27raw 0.14 / 0.25
Teasing
z +0.21 / -0.04raw 0.12 / -0.02
Interest
z +0.20 / +0.15raw 1.77 / 1.73
Contempt
z +0.16 / +0.10raw 0.08 / 0.05

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Newsreader style (S_NEWS)
z +4.69 / +4.69raw 3.07 / 3.07
Formal style (S_FORM)
z +2.76 / +2.82raw 4.01 / 4.07
Teacher or didactic style (S_TECH)
z +1.91 / +2.42raw 2.98 / 3.61
Speech structure (STRU)
z +1.90 / +2.15raw 3.95 / 4.20
Authoritative style (S_AUTH)
z +1.99 / +1.92raw 3.16 / 3.09
Mixed resonance (R_MIXD)
z +1.31 / +1.70raw 2.54 / 2.87
Fullness (FULL)
z +1.02 / +1.65raw 2.68 / 3.25

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z -1.04 / -1.33raw 0.64 / 0.18
Vocal-burst blend (0–10)z -1.59 / -1.59raw 2.33 / 2.33
Content enjoyment (CE)z +1.09 / +0.74raw 5.91 / 5.55
Production quality (PQ)z +0.80 / +0.37raw 7.75 / 7.37
Speech signal quality, calibratedz +0.27 / +0.36raw 3.62 / 3.67
Background quality, calibratedz +1.05 / +0.94raw 4.37 / 4.26
Overall quality, calibratedz +1.05 / +1.04raw 3.48 / 3.47
P.808 MOS estimatez +0.93 / +1.01raw 3.88 / 3.92
score content enjoymentz +0.70 / +0.87raw 4.72 / 4.77
score overall qualityz +0.75 / +1.04raw 3.03 / 3.18
Background Noisez -1.91 / -2.05raw 0.30 / 0.27
Tempo (TEMP)z +1.00 / +1.03raw 3.31 / 3.34
Arousal (AROU)z +0.29 / +0.25raw 2.69 / 2.65
Valence (VALN)z +0.50 / +0.33raw 2.89 / 2.69
Recording quality (RCQL)z +1.55 / +1.53raw 4.13 / 4.11
Background noise (BKGN)z +1.00 / +1.04raw 3.47 / 3.49
talking speedz +1.67 / +1.39raw 15.29 / 14.26
Characters per second (CPS): 6.20 raw · z -0.75

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8580.863
Identity · 250D0.8000.783

Vocal-burst regions

Tiny
Base
0 s8.4 s

Verified burst intervals are unavailable for this clip.

Tiny: 0 regions, count-head estimate 0.03; Base: 0 regions, count-head estimate 0.00.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.049+0.070.041+0.060.0070.570
1Amusementemo_Amusement0.061+0.070.027+0.03-0.0010.885
2Angeremo_Anger0.094+0.150.092+0.140.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise-0.020-0.04-0.033-0.070.0050.572
4Aweemo_Awe0.009+0.020.017+0.040.0000.427
5Bitternessemo_Bitterness0.057+0.100.024+0.030.0130.415
6Concentrationemo_Concentration1.704+1.441.641+1.360.6340.741
7Confusionemo_Confusion-0.016-0.03-0.003-0.010.0010.540
8Contemplationemo_Contemplation0.399+0.530.326+0.410.0820.598
9Contemptemo_Contempt0.076+0.160.046+0.10-0.0000.464
10Contentmentemo_Contentment0.137+0.270.226+0.440.0000.514
11Disappointmentemo_Disappointment0.005+0.010.060+0.090.0000.700
12Disgustemo_Disgust0.017+0.020.040+0.070.0070.489
13Distressemo_Distress-0.001-0.000.095+0.120.0000.821
14Doubtemo_Doubt0.006+0.000.004-0.000.0060.466
15Elationemo_Elation-0.020-0.040.009-0.010.0120.728
16Embarrassmentemo_Embarrassment-0.042-0.20-0.022-0.150.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.144+0.150.252+0.270.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.023-0.010.031+0.010.0280.539
19Fearemo_Fear-0.032-0.08-0.034-0.080.0150.610
20Helplessnessemo_Helplessness0.035+0.060.063+0.10-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.098-0.030.068-0.060.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.036-0.160.020-0.180.1490.705
23Infatuationemo_Infatuation-0.031-0.03-0.000+0.03-0.0150.533
24Interestemo_Interest1.767+0.201.728+0.151.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness-0.012+0.02-0.013+0.02-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.068+0.020.079+0.040.0590.442
27Longingemo_Longing0.000-0.140.010-0.120.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.102+0.090.031-0.040.0550.534
29Painemo_Pain0.022+0.04-0.033-0.05-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.030+0.020.006-0.020.0170.614
31Prideemo_Pride0.062+0.120.078+0.150.0020.509
32Reliefemo_Relief0.026-0.010.031-0.000.0330.681
33Sadnessemo_Sadness0.038+0.050.017+0.020.0000.759
34Sexual Lustemo_Sexual_Lust0.051+0.04-0.027-0.110.0300.494
35Shameemo_Shame0.011-0.000.048+0.070.0130.510
36Sournessemo_Sourness0.036+0.090.001+0.01-0.0050.439
37Teasingemo_Teasing0.117+0.21-0.020-0.040.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.031-0.010.099+0.080.0410.697
39Triumphemo_Triumph-0.006+0.00-0.029-0.04-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg2.891-0.082.776-0.172.9841.238
41Arousal (AROU)vn_AROU_reg2.693+0.292.648+0.252.3591.147
42Arousal shift (ARSH)vn_ARSH_reg2.533-0.052.762+0.212.5800.867
43Vocal attack (ATCK)vn_ATCK_reg2.025-0.111.747-0.322.1621.295
44Background noise (BKGN)vn_BKGN_reg3.468+1.003.494+1.042.7030.761
45Brightness (BRGT)vn_BRGT_reg2.078+0.112.128+0.171.9910.813
46Speech chunking (CHNK)vn_CHNK_reg2.400+0.452.435+0.471.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg3.328+1.393.396+1.462.0180.945
48Cognitive load (COGL)vn_COGL_reg1.087-0.950.943-1.092.0941.061
49Dynamic arc (DARC)vn_DARC_reg2.006+0.281.746+0.091.6151.380
50Disfluency (DFLU)vn_DFLU_reg0.545-1.640.490-1.692.4881.184
51Emphasis (EMPH)vn_EMPH_reg2.482-0.082.466-0.092.6071.524
52Esthetics (ESTH)vn_ESTH_reg2.710+0.392.653+0.322.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.002-0.26-0.042-0.400.0860.322
54Focus (FOCS)vn_FOCS_reg2.573+0.212.770+0.392.3441.102
55Fullness (FULL)vn_FULL_reg2.683+1.023.249+1.651.7780.889
56Perceived gender (GEND)vn_GEND_reg4.009-0.034.391+0.214.0551.588
57Harmonicity (HARM)vn_HARM_reg3.196+0.862.847+0.442.4800.836
58Metallic character (METL)vn_METL_reg1.702+0.081.687+0.071.6210.958
59Pitch range (RANG)vn_RANG_reg2.723-0.212.787-0.142.9140.897
60Recording quality (RCQL)vn_RCQL_reg4.130+1.554.109+1.532.7540.888
61Vocal register (REGS)vn_REGS_reg1.086-0.091.101-0.081.1981.268
62Respiration (RESP)vn_RESP_reg1.535-1.081.510-1.112.6801.056
63Roughness (ROUG)vn_ROUG_reg1.637-0.451.437-0.672.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg2.638+0.602.930+0.942.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg2.562+0.502.546+0.482.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg2.169+0.261.940+0.031.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg2.540+1.312.867+1.701.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg2.122-0.642.041-0.722.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg2.331+0.202.348+0.222.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg2.235+0.202.145+0.092.0700.803
71Smoothness (SMTH)vn_SMTH_reg3.686+1.413.773+1.492.1431.093
72Stance (STNC)vn_STNC_reg2.932+0.262.532-0.092.6331.137
73Speech structure (STRU)vn_STRU_reg3.955+1.904.202+2.152.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg0.677-0.490.478-0.631.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg3.159+1.993.094+1.921.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg0.417-0.820.310-0.891.5601.401
77Casual style (S_CASU)vn_S_CASU_reg2.006-1.401.777-1.583.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg1.483-0.841.955-0.482.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg2.359-0.322.103-0.542.7211.148
80Formal style (S_FORM)vn_S_FORM_reg4.011+2.764.074+2.821.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg3.759+0.733.630+0.632.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg2.742+0.992.708+0.971.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg3.070+4.693.070+4.690.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg1.626-0.551.799-0.422.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg0.323-0.690.482-0.571.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg1.659-0.601.808-0.492.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg2.984+1.913.612+2.420.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg1.024-0.351.186-0.251.5971.618
89Tempo (TEMP)vn_TEMP_reg3.306+1.003.344+1.032.1171.190
90Tension (TENS)vn_TENS_reg1.903+0.401.615+0.071.5590.852
91Valence (VALN)vn_VALN_reg2.888+0.502.691+0.332.3281.118
92Valence shift (VALS)vn_VALS_reg2.965+0.422.869+0.332.5181.079
93Velocity flux (VFLX)vn_VFLX_reg2.544+0.242.492+0.202.2561.208
94Volatility (VOLT)vn_VOLT_reg1.382-1.361.314-1.442.5860.886
95Vulnerability (VULN)vn_VULN_reg1.939-0.721.595-1.072.6480.984
96Warmth (WARM)vn_WARM_reg2.900+0.353.006+0.472.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_60.641-1.040.177-1.332.3321.621
98Vocal-burst blend (0–10)blend_0_102.333-1.592.333-1.596.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.350-2.210.283-2.401.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.249-0.013.262+0.043.2500.293
101Arousaleiv_extra:Arousal1.022-0.610.997-0.671.2890.435
102Authenticityeiv_extra:Authenticity2.708+0.272.669+0.012.6670.153
103Background Noiseeiv_extra:Background_Noise0.303-1.910.266-2.050.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant0.919-0.400.961-0.261.0400.303
105Gendereiv_extra:Gender-0.717-0.10-1.060-0.44-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.889+0.251.887+0.241.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive1.382-1.021.391-1.001.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.544+1.202.554+1.242.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous0.840-0.260.870-0.170.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.151-0.410.104-0.550.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.307-0.210.251-0.390.3760.324
112Valenceeiv_extra:Valence0.017+0.03-0.007-0.030.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached2.416+1.612.406+1.581.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold0.073+0.070.016-0.120.0510.300
115score background qualityeiv_extra:score_background_quality3.743+0.643.832+0.773.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.719+0.704.771+0.874.5070.303
117score overall qualityeiv_extra:score_overall_quality3.030+0.753.179+1.042.6520.507
118score speech qualityeiv_extra:score_speech_quality1.921+0.781.931+0.851.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.911+1.095.550+0.744.7991.018
120Content usefulness (CU)audiobox:CU6.530+0.886.183+0.535.6491.001
121Production complexity (PC)audiobox:PC1.511-0.511.612-0.442.3311.617
122Production quality (PQ)audiobox:PQ7.747+0.807.366+0.377.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw4.152+0.384.212+0.473.9000.669
124Background quality, rawdnsmos:BAK_raw4.667+1.304.553+1.193.2111.124
125Overall quality, rawdnsmos:OVRL_raw4.043+1.184.043+1.183.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.624+0.273.667+0.363.4880.501
127Background quality, calibrateddnsmos:BAK4.365+1.054.264+0.943.3890.928
128Overall quality, calibrateddnsmos:OVRL3.477+1.053.473+1.042.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.880+0.933.923+1.013.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.033+0.050.002+0.000.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.106-1.520.253-1.250.9470.554
132Agevoiceclap_attribute:Age3.071-0.293.351+0.223.2290.547
133Amusementvoiceclap_attribute:Amusement-0.047-0.77-0.056-0.780.3920.572
134Angervoiceclap_attribute:Anger0.236-1.490.232-1.500.9920.506
135Arousalvoiceclap_attribute:Arousal0.745-1.080.842-0.931.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.262-0.510.331-0.370.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.690-0.482.670-0.602.7700.167
138Awevoiceclap_attribute:Awe0.324-0.830.227-1.200.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.255-1.180.199-1.400.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.253-1.440.124-1.990.5920.236
141Concentrationvoiceclap_attribute:Concentration1.606+0.741.542+0.621.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.058-0.030.934-0.291.0690.468
143Confusionvoiceclap_attribute:Confusion0.273-1.110.197-1.340.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.587-0.480.281-1.230.7800.406
145Contemptvoiceclap_attribute:Contempt0.104-1.790.035-1.990.7090.339
146Contentmentvoiceclap_attribute:Contentment0.237-1.140.359-0.750.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.291-1.610.232-1.810.7600.291
148Disgustvoiceclap_attribute:Disgust0.275-1.690.135-2.370.6250.207
149Distressvoiceclap_attribute:Distress0.155-2.210.068-2.460.9270.350
150Doubtvoiceclap_attribute:Doubt0.201-1.780.105-2.280.5350.188
151Elationvoiceclap_attribute:Elation0.174-0.920.336-0.510.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.098-0.660.137-0.540.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.751+0.330.773+0.410.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.087-1.440.014-1.720.4620.261
155Fearvoiceclap_attribute:Fear0.304-1.480.175-2.090.6180.212
156Gendervoiceclap_attribute:Gender-0.441-0.29-1.006-0.85-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.363-1.100.164-1.840.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.783-0.151.762-0.211.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.395-0.780.501-0.520.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability0.087-1.780.011-1.931.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.316-0.880.339-0.810.6080.332
162Interestvoiceclap_attribute:Interest1.742+0.331.842+0.551.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.144-0.770.182-0.690.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.399-0.500.186-1.380.5230.245
165Longingvoiceclap_attribute:Longing0.121-1.90-0.014-2.510.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.214-1.790.089-2.150.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive1.423-0.591.458-0.541.7750.594
168Painvoiceclap_attribute:Pain0.076-1.400.060-1.470.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy-0.072-1.910.053-1.500.5030.301
170Pridevoiceclap_attribute:Pride0.565-0.620.579-0.580.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.679+1.122.740+1.302.3130.328
172Reliefvoiceclap_attribute:Relief0.299-1.750.416-1.471.0250.415
173Sadnessvoiceclap_attribute:Sadness0.276-1.520.171-1.860.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.776-0.050.689-0.200.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.271-0.870.190-1.060.6350.419
176Shamevoiceclap_attribute:Shame0.163-1.280.207-1.130.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.247+0.240.061-0.270.1600.360
178Sournessvoiceclap_attribute:Sourness0.337-1.270.228-1.640.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.329-0.820.432-0.530.6230.359
180Teasingvoiceclap_attribute:Teasing-0.091-1.670.087-1.200.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude0.085-2.450.403-1.851.3840.529
182Triumphvoiceclap_attribute:Triumph0.344-1.030.436-0.780.7160.359
183Valencevoiceclap_attribute:Valence0.256+0.410.211+0.34-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached2.387+1.352.369+1.311.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold-0.051+0.04-0.119-0.13-0.0660.400
186durationvoiceclap_attribute:duration10.188-0.119.767-0.1810.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.725+0.963.885+1.283.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.868+1.444.947+1.984.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality3.035+0.973.261+1.522.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality2.010+1.722.017+1.821.8910.069
191talking speedvoiceclap_attribute:talking_speed15.286+1.6714.260+1.399.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.28890.27320.1665
1-0.0401-0.0138-0.0842
20.07220.12260.0593
30.15830.13240.1292
40.09260.09570.1183
5-0.0688-0.1376-0.1072
60.00570.03830.0927
7-0.0066-0.03540.0679
8-0.0526-0.0714-0.0160
90.10600.10660.1238
100.21970.21830.2725
11-0.1529-0.1811-0.0991
12-0.0140-0.0957-0.0086
13-0.00240.04400.0712
14-0.0319-0.0241-0.0431
15-0.0221-0.0735-0.0759
16-0.0373-0.02380.0135
17-0.1006-0.1102-0.0881
180.08110.07090.0576
190.09800.08070.0572
20-0.1175-0.0962-0.0848
210.11680.03880.0561
22-0.0012-0.0091-0.0545
23-0.0672-0.0921-0.0367
24-0.0730-0.0807-0.0684
25-0.0903-0.0859-0.0744
26-0.0232-0.0388-0.0233
27-0.0711-0.0776-0.0790
28-0.1291-0.1247-0.0933
29-0.0722-0.0484-0.0726
30-0.0144-0.0349-0.0503
310.11990.13560.1170
320.08070.08070.1716
330.0411-0.0022-0.0732
34-0.0557-0.03910.0294
35-0.0807-0.0968-0.0979
360.01530.04400.0413
37-0.0492-0.0854-0.1003
380.08910.09050.0493
39-0.0271-0.0235-0.0449
400.07990.00260.0511
41-0.0472-0.0092-0.1218
42-0.06030.01570.0738
43-0.0811-0.0983-0.0872
44-0.0607-0.0942-0.0814
450.0307-0.01130.0455
46-0.1037-0.1014-0.0799
47-0.0515-0.02210.0469
480.03820.01220.0061
49-0.1260-0.1180-0.0856
500.10370.01160.1431
51-0.0218-0.0492-0.0155
52-0.0780-0.0781-0.0963
53-0.0937-0.1118-0.0864
54-0.1168-0.1376-0.0948
55-0.0853-0.1035-0.0815
560.12600.11640.1971
57-0.0569-0.0952-0.0930
580.02770.09520.1351
59-0.0095-0.06830.0001
600.07910.0486-0.0320
61-0.0390-0.0259-0.0663
62-0.00190.0512-0.0024
63-0.0378-0.0592-0.0805
640.05070.03830.0319
65-0.0960-0.0435-0.0691
660.21200.21420.1992
670.18980.16560.2091
68-0.0145-0.0261-0.0126
69-0.0715-0.0109-0.0375
70-0.0350-0.0551-0.0520
71-0.0499-0.0828-0.0115
720.09600.13660.1011
730.11290.09420.0698
74-0.0903-0.0719-0.0926
750.13680.11690.1111
76-0.0003-0.04500.0365
77-0.0319-0.02880.0341
780.03440.07190.0564
79-0.1076-0.0962-0.0735
800.11370.05350.0580
810.05380.07090.1003
82-0.0199-0.0295-0.0811
830.08110.12370.1259
84-0.03840.0059-0.0322
85-0.0795-0.0582-0.0900
860.12980.09680.0637
87-0.1122-0.1324-0.0881
880.17820.11490.1313
89-0.0061-0.0056-0.0367
90-0.0448-0.0229-0.0919
910.05300.08640.1031
92-0.0634-0.0213-0.0452
93-0.0233-0.0559-0.0239
94-0.0089-0.0541-0.0234
95-0.0968-0.1138-0.0910
96-0.1099-0.1268-0.0504
97-0.0584-0.0057-0.0592
98-0.0653-0.1190-0.0513
990.09300.12160.1611
1000.11370.08380.1110
101-0.0279-0.0797-0.0819
1020.05840.03270.0690
103-0.05030.0370-0.0420
104-0.1037-0.0735-0.0829
105-0.0983-0.0874-0.0810
106-0.0860-0.0962-0.0681
1070.15520.15310.1881
108-0.0319-0.00910.0166
1090.00570.05950.0777
1100.03150.02540.1362
111-0.0822-0.0585-0.0743
112-0.0968-0.0823-0.0835
1130.02360.02730.0352
1140.04360.04890.0903
115-0.0392-0.0504-0.0803
116-0.0957-0.0807-0.0726
117-0.0005-0.04160.0393
118-0.00860.01090.0417
119-0.1168-0.0885-0.0819
1200.16060.14590.1157
121-0.0864-0.0365-0.0106
122-0.0757-0.0688-0.0375
1230.0038-0.0383-0.0729
1240.06380.0823-0.0242
125-0.1283-0.1019-0.0845
126-0.0699-0.0900-0.0731
127-0.1575-0.1635-0.0900
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
00.0061-0.01670.0598
1-0.0381-0.0461-0.0403
20.02380.02220.0194
3-0.0507-0.0975-0.0638
4-0.0049-0.0288-0.0379
5-0.0054-0.0328-0.0435
6-0.0173-0.0223-0.0610
70.24990.21050.1876
80.02070.0020-0.0287
90.08650.12480.0808
100.00600.01960.0108
11-0.1094-0.1004-0.0563
12-0.0236-0.0392-0.0050
130.17820.17790.1305
14-0.0139-0.0039-0.0594
150.09280.09060.0225
160.0230-0.0018-0.0673
17-0.0375-0.0363-0.0084
180.10200.15990.1299
19-0.0917-0.0681-0.0660
20-0.0416-0.0068-0.0417
21-0.0525-0.0345-0.0386
22-0.0348-0.0455-0.0432
23-0.0514-0.0604-0.0403
24-0.0569-0.0451-0.0356
25-0.01360.00790.0055
260.10720.13050.0566
27-0.0234-0.0298-0.0338
280.00800.02690.0272
290.0466-0.01040.1437
30-0.0595-0.0636-0.0362
31-0.0082-0.0485-0.0073
320.05290.0298-0.0269
330.06400.00520.0634
340.02360.02090.0004
35-0.0090-0.0084-0.0238
36-0.0342-0.0624-0.0385
37-0.0079-0.0133-0.0288
38-0.0046-0.03240.0205
39-0.0464-0.06280.0160
40-0.0440-0.0218-0.0102
41-0.0384-0.0308-0.0478
420.02980.01750.0453
430.06290.09060.1249
440.05920.06890.0555
45-0.01720.00680.0099
460.02340.06450.0878
47-0.0003-0.0158-0.0147
48-0.0507-0.0290-0.0444
49-0.0983-0.1040-0.0464
500.00000.0608-0.0369
510.0022-0.04340.0016
52-0.0353-0.0543-0.0411
53-0.0052-0.0312-0.0403
540.05290.02840.0261
550.07690.03390.0085
560.0040-0.02590.0330
570.08360.07060.0515
58-0.0625-0.0530-0.0399
590.04810.05590.0164
600.09020.06160.0091
610.13090.15670.2303
620.0081-0.0234-0.0366
630.00500.02130.0133
64-0.0580-0.0975-0.0464
65-0.0290-0.0418-0.0014
66-0.0442-0.0649-0.0524
67-0.0714-0.0379-0.0438
680.05840.07380.0641
690.0042-0.0483-0.0768
70-0.0836-0.0632-0.0604
71-0.0210-0.0483-0.0436
72-0.0503-0.0443-0.0461
730.01990.0228-0.0357
74-0.0573-0.0742-0.0449
75-0.0168-0.02710.0340
760.0001-0.03790.0207
77-0.0201-0.0465-0.0218
78-0.0447-0.0240-0.0607
790.01380.0062-0.0338
80-0.0525-0.1044-0.0442
810.05250.0088-0.0004
82-0.0566-0.0787-0.0399
83-0.0085-0.01830.0089
84-0.0073-0.03200.0171
85-0.0762-0.0889-0.0445
86-0.0283-0.0191-0.0397
87-0.0484-0.0410-0.0355
880.09980.13300.0958
890.01190.0253-0.0402
900.16270.10440.1299
910.21300.23990.1704
92-0.0817-0.0685-0.0384
930.02000.0669-0.0287
940.05510.02230.0390
95-0.0046-0.0373-0.0464
960.03840.0166-0.0123
97-0.0558-0.0543-0.0374
980.04160.02190.0462
99-0.0189-0.0310-0.0556
1000.09980.06450.0477
101-0.0743-0.0849-0.0420
1020.0055-0.01780.0695
1030.03420.00990.0343
104-0.0717-0.0718-0.0579
105-0.0858-0.0475-0.0578
106-0.0244-0.0261-0.0342
1070.12790.0067-0.0398
108-0.0129-0.0441-0.0331
1090.08470.07910.0660
110-0.0858-0.0694-0.0529
1110.09240.04180.0322
112-0.0222-0.0151-0.0352
113-0.0272-0.05710.0753
1140.02680.01210.0016
115-0.0089-0.0023-0.0315
1160.0137-0.03040.0549
1170.04730.04100.0142
118-0.0065-0.0371-0.0033
1190.09690.09790.1135
1200.01610.0202-0.0096
1210.04140.02000.0798
122-0.0349-0.0383-0.0420
1230.0043-0.0148-0.0459
124-0.0470-0.0632-0.0443
1250.0211-0.00470.0861
126-0.0390-0.0608-0.0645
127-0.0161-0.05340.0703
1280.0149-0.00380.0282
1290.0048-0.03690.0239
130-0.0569-0.0337-0.0446
131-0.0407-0.0430-0.0389
1320.00340.0365-0.0070
1330.0212-0.0171-0.0355
1340.02640.04080.0144
135-0.0470-0.0608-0.0447
136-0.0518-0.0339-0.0450
1370.0044-0.0070-0.0413
138-0.0442-0.0357-0.0392
1390.07650.05470.1674
140-0.0242-0.0555-0.0524
1410.04050.0381-0.0440
142-0.0036-0.0335-0.0241
143-0.0396-0.0057-0.0272
1440.00950.03260.0650
145-0.0706-0.0496-0.0516
146-0.0462-0.0698-0.0140
147-0.0270-0.0477-0.0395
148-0.0521-0.0447-0.0490
1490.08280.01760.0330
1500.08280.01140.0358
1510.00670.03790.0025
152-0.0366-0.0575-0.0088
153-0.0157-0.0010-0.0407
1540.0518-0.02750.0180
1550.04730.01180.1135
156-0.0181-0.0168-0.0445
1570.12200.12650.1105
158-0.1383-0.1289-0.0553
159-0.0802-0.0738-0.0520
160-0.0234-0.0575-0.0151
161-0.01040.00220.0288
162-0.0033-0.0379-0.0159
1630.02050.03530.0987
164-0.00390.01450.0013
1650.03220.06120.0316
166-0.0069-0.0579-0.0475
1670.06030.0632-0.0111
168-0.0684-0.0475-0.0101
1690.0462-0.0107-0.0188
170-0.0401-0.0563-0.0416
171-0.0606-0.0530-0.0391
172-0.0976-0.0702-0.0515
1730.0274-0.02110.0285
174-0.0499-0.0359-0.0456
175-0.0381-0.0108-0.0673
1760.12420.05200.1193
177-0.0359-0.0447-0.0366
1780.22030.15750.1609
179-0.0105-0.0168-0.0137
1800.0880-0.00620.0561
1810.03140.03280.0508
1820.00220.00850.0249
183-0.0346-0.0151-0.0897
184-0.00590.0107-0.0129
1850.03480.0187-0.0226
186-0.0847-0.0653-0.0536
1870.16640.14360.2119
188-0.0137-0.0187-0.0417
1890.05320.04900.1487
190-0.0106-0.0019-0.0511
1910.02810.0000-0.0113
192-0.0383-0.04810.0379
1930.07950.04670.0006
1940.0384-0.01420.0141
1950.0063-0.0290-0.0210
196-0.01310.0012-0.0432
197-0.0262-0.0281-0.0622
198-0.0351-0.0512-0.0429
199-0.0240-0.0665-0.0093
200-0.0451-0.0587-0.0410
2010.12720.07510.0413
202-0.0390-0.0127-0.0014
203-0.00860.01570.0074
2040.10200.12730.1176
205-0.0558-0.0543-0.0715
206-0.0151-0.0636-0.0458
207-0.0335-0.0371-0.0525
208-0.0529-0.0475-0.0461
2090.02960.02600.0328
2100.03880.05870.0472
2110.10650.10120.1622
2120.02350.0479-0.0532
213-0.00050.0051-0.0131
2140.13980.08770.1527
2150.08910.15420.0923
2160.06800.05750.0813
2170.08430.08240.0703
2180.06620.10930.0900
219-0.0758-0.0971-0.0578
220-0.1139-0.1134-0.0490
2210.09830.14030.1515
2220.0165-0.0024-0.0068
2230.04920.0263-0.0088
224-0.0325-0.0386-0.0406
225-0.0349-0.0369-0.0370
2260.16420.14770.1511
2270.04400.08400.1123
228-0.0658-0.0730-0.0719
2290.02720.02960.0495
2300.06430.05430.0268
231-0.0521-0.0869-0.0615
232-0.01870.00730.0620
2330.03030.04610.0089
234-0.0294-0.0512-0.0377
2350.17970.20320.2115
236-0.0566-0.0526-0.0381
2370.06250.04220.1198
2380.05400.06400.0844
239-0.0773-0.1191-0.0459
240-0.0636-0.0475-0.0453
2410.0008-0.0426-0.0098
242-0.0654-0.0500-0.0654
243-0.0373-0.0604-0.0205
2440.08870.07550.0495
2450.06580.05020.0276
246-0.0219-0.0432-0.0360
247-0.02230.0105-0.0374
248-0.0357-0.0343-0.0107
2490.01380.03610.0775
Source: Emolia / Emilia source archive · SHA-256 547e07f8272cd3ca…
e12ChineseEmolia / Emilia

Emolia ZH · ZH_B00000_S00120_W000001

Natural speech · Original speech · 13.8 s

从今天开始,为您分享本书的第七章强大原型案例深入解读,让我们再重复一遍推广定律,有效推广定律。

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Concentration
z +1.47 / +1.52raw 1.73 / 1.76
Contentment
z +1.14 / +1.38raw 0.59 / 0.71
Contemplation
z +0.85 / +0.86raw 0.59 / 0.60
Doubt
z +0.41 / +0.21raw 0.20 / 0.10
Confusion
z +0.40 / +0.20raw 0.22 / 0.11
Hope Enthusiasm Optimism
z +0.38 / +0.23raw 0.40 / 0.29
Sourness
z +0.26 / +0.01raw 0.11 / -0.00

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Newsreader style (S_NEWS)
z +4.50 / +4.11raw 2.97 / 2.74
Teacher or didactic style (S_TECH)
z +3.18 / +3.12raw 4.54 / 4.47
Formal style (S_FORM)
z +2.91 / +2.48raw 4.17 / 3.73
Vocal register (REGS)
z +2.34 / +2.32raw 4.17 / 4.14
Articulation clarity (CLRT)
z +2.05 / +1.70raw 3.95 / 3.62
Whispered talking style (S_WHIS)
z +1.56 / +1.87raw 4.13 / 4.63
Authoritative style (S_AUTH)
z +1.86 / +1.73raw 3.03 / 2.89

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z -1.10 / -1.08raw 0.54 / 0.58
Vocal-burst blend (0–10)z -1.55 / -1.49raw 2.44 / 2.61
Content enjoyment (CE)z +0.91 / +0.82raw 5.72 / 5.63
Production quality (PQ)z +0.68 / +0.52raw 7.64 / 7.50
Speech signal quality, calibratedz +0.19 / +0.13raw 3.59 / 3.55
Background quality, calibratedz +0.90 / +0.90raw 4.22 / 4.23
Overall quality, calibratedz +0.86 / +0.81raw 3.36 / 3.33
P.808 MOS estimatez +1.05 / +1.27raw 3.94 / 4.05
score content enjoymentz +0.27 / +0.31raw 4.59 / 4.60
score overall qualityz +0.93 / +1.07raw 3.12 / 3.19
Background Noisez -1.09 / -1.35raw 0.52 / 0.45
Tempo (TEMP)z +0.24 / +0.30raw 2.41 / 2.47
Arousal (AROU)z -0.41 / -0.51raw 1.89 / 1.78
Valence (VALN)z +0.54 / +0.26raw 2.94 / 2.62
Recording quality (RCQL)z +1.48 / +1.45raw 4.07 / 4.04
Background noise (BKGN)z +1.42 / +1.56raw 3.79 / 3.89
talking speedz -0.22 / +0.08raw 8.21 / 9.33
Characters per second (CPS): 3.40 raw · z -1.50

Characters in source transcript divided by audio seconds; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8370.828
Identity · 250D0.8180.852

Vocal-burst regions

Tiny
Base
0 s13.8 s

Verified burst intervals are unavailable for this clip.

Tiny: 0 regions, count-head estimate 0.03; Base: 0 regions, count-head estimate 0.01.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.053+0.08-0.005-0.020.0070.570
1Amusementemo_Amusement-0.004-0.00-0.081-0.09-0.0010.885
2Angeremo_Anger0.009+0.01-0.018-0.040.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.003-0.00-0.033-0.070.0050.572
4Aweemo_Awe0.034+0.08-0.008-0.020.0000.427
5Bitternessemo_Bitterness0.000-0.030.041+0.070.0130.415
6Concentrationemo_Concentration1.727+1.471.761+1.520.6340.741
7Confusionemo_Confusion0.216+0.400.110+0.200.0010.540
8Contemplationemo_Contemplation0.593+0.850.598+0.860.0820.598
9Contemptemo_Contempt0.024+0.050.040+0.09-0.0000.464
10Contentmentemo_Contentment0.588+1.140.710+1.380.0000.514
11Disappointmentemo_Disappointment-0.055-0.080.035+0.050.0000.700
12Disgustemo_Disgust0.075+0.140.054+0.100.0070.489
13Distressemo_Distress-0.055-0.070.063+0.080.0000.821
14Doubtemo_Doubt0.199+0.410.102+0.210.0060.466
15Elationemo_Elation0.111+0.14-0.098-0.150.0120.728
16Embarrassmentemo_Embarrassment-0.068-0.26-0.032-0.170.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.120+0.130.165+0.180.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.026-0.000.013-0.030.0280.539
19Fearemo_Fear-0.099-0.19-0.068-0.130.0150.610
20Helplessnessemo_Helplessness0.035+0.06-0.053-0.07-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.404+0.380.290+0.230.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.055-0.130.049-0.140.1490.705
23Infatuationemo_Infatuation-0.046-0.060.011+0.05-0.0150.533
24Interestemo_Interest1.815+0.251.781+0.211.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness0.044+0.12-0.039-0.03-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.076+0.040.049-0.020.0590.442
27Longingemo_Longing-0.034-0.19-0.019-0.170.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.051-0.010.021-0.060.0550.534
29Painemo_Pain0.028+0.05-0.022-0.03-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy-0.020-0.06-0.077-0.150.0170.614
31Prideemo_Pride0.082+0.160.045+0.080.0020.509
32Reliefemo_Relief0.040+0.010.010-0.030.0330.681
33Sadnessemo_Sadness-0.049-0.06-0.010-0.010.0000.759
34Sexual Lustemo_Sexual_Lust0.024-0.01-0.045-0.150.0300.494
35Shameemo_Shame0.002-0.020.020+0.010.0130.510
36Sournessemo_Sourness0.110+0.26-0.003+0.01-0.0050.439
37Teasingemo_Teasing0.128+0.24-0.083-0.150.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.079+0.050.093+0.070.0410.697
39Triumphemo_Triumph-0.001+0.01-0.016-0.02-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg3.873+0.724.266+1.042.9841.238
41Arousal (AROU)vn_AROU_reg1.889-0.411.777-0.512.3591.147
42Arousal shift (ARSH)vn_ARSH_reg2.054-0.611.893-0.792.5800.867
43Vocal attack (ATCK)vn_ATCK_reg1.826-0.261.784-0.292.1621.295
44Background noise (BKGN)vn_BKGN_reg3.787+1.423.893+1.562.7030.761
45Brightness (BRGT)vn_BRGT_reg1.585-0.501.677-0.391.9910.813
46Speech chunking (CHNK)vn_CHNK_reg1.397-0.361.259-0.471.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg3.951+2.053.621+1.702.0180.945
48Cognitive load (COGL)vn_COGL_reg2.730+0.602.319+0.212.0941.061
49Dynamic arc (DARC)vn_DARC_reg1.724+0.081.269-0.251.6151.380
50Disfluency (DFLU)vn_DFLU_reg0.840-1.390.711-1.502.4881.184
51Emphasis (EMPH)vn_EMPH_reg2.315-0.192.475-0.092.6071.524
52Esthetics (ESTH)vn_ESTH_reg2.681+0.362.854+0.582.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.076-0.030.325+0.740.0860.322
54Focus (FOCS)vn_FOCS_reg1.140-1.091.413-0.842.3441.102
55Fullness (FULL)vn_FULL_reg1.946+0.191.793+0.021.7780.889
56Perceived gender (GEND)vn_GEND_reg0.559-2.200.982-1.934.0551.588
57Harmonicity (HARM)vn_HARM_reg2.469-0.012.431-0.062.4800.836
58Metallic character (METL)vn_METL_reg2.073+0.472.122+0.521.6210.958
59Pitch range (RANG)vn_RANG_reg2.380-0.602.728-0.212.9140.897
60Recording quality (RCQL)vn_RCQL_reg4.067+1.484.039+1.452.7540.888
61Vocal register (REGS)vn_REGS_reg4.170+2.344.141+2.321.1981.268
62Respiration (RESP)vn_RESP_reg2.308-0.352.491-0.182.6801.056
63Roughness (ROUG)vn_ROUG_reg1.875-0.182.057+0.032.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg2.305+0.202.331+0.232.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg2.312+0.202.372+0.272.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg2.403+0.492.229+0.321.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg2.658+1.452.919+1.761.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg3.040+0.243.205+0.402.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg2.304+0.162.086-0.132.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg2.386+0.392.095+0.032.0700.803
71Smoothness (SMTH)vn_SMTH_reg3.122+0.903.435+1.182.1431.093
72Stance (STNC)vn_STNC_reg2.710+0.072.496-0.122.6331.137
73Speech structure (STRU)vn_STRU_reg3.662+1.593.770+1.702.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg3.213+1.313.332+1.391.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg3.030+1.862.892+1.731.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg2.031+0.341.798+0.171.5601.401
77Casual style (S_CASU)vn_S_CASU_reg1.488-1.801.617-1.703.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg1.809-0.602.021-0.432.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg1.871-0.741.604-0.972.7211.148
80Formal style (S_FORM)vn_S_FORM_reg4.168+2.913.727+2.481.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg3.988+0.914.247+1.112.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg3.213+1.323.347+1.411.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg2.967+4.502.743+4.110.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg2.244-0.081.893-0.352.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg-0.101-1.02-0.161-1.071.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg2.076-0.302.285-0.142.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg4.544+3.184.468+3.120.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg4.127+1.564.627+1.871.5971.618
89Tempo (TEMP)vn_TEMP_reg2.408+0.242.471+0.302.1171.190
90Tension (TENS)vn_TENS_reg1.597+0.051.139-0.491.5590.852
91Valence (VALN)vn_VALN_reg2.936+0.542.623+0.262.3281.118
92Valence shift (VALS)vn_VALS_reg2.618+0.092.547+0.032.5181.079
93Velocity flux (VFLX)vn_VFLX_reg2.159-0.082.077-0.152.2561.208
94Volatility (VOLT)vn_VOLT_reg0.744-2.080.812-2.002.5860.886
95Vulnerability (VULN)vn_VULN_reg2.148-0.512.191-0.472.6480.984
96Warmth (WARM)vn_WARM_reg2.975+0.443.340+0.842.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_60.541-1.100.579-1.082.3321.621
98Vocal-burst blend (0–10)blend_0_102.440-1.552.610-1.496.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.355-2.190.376-2.131.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.308+0.203.336+0.293.2500.293
101Arousaleiv_extra:Arousal1.017-0.621.016-0.631.2890.435
102Authenticityeiv_extra:Authenticity2.744+0.512.798+0.862.6670.153
103Background Noiseeiv_extra:Background_Noise0.518-1.090.451-1.350.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant0.961-0.261.031-0.031.0400.303
105Gendereiv_extra:Gender1.333+1.961.177+1.80-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.762-0.261.690-0.551.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive1.662-0.421.687-0.371.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.497+1.062.525+1.152.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous0.701-0.670.795-0.400.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.565+0.890.620+1.060.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.332-0.130.139-0.730.3760.324
112Valenceeiv_extra:Valence-0.021-0.070.069+0.170.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.820+0.241.763+0.111.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold0.063+0.040.116+0.220.0510.300
115score background qualityeiv_extra:score_background_quality3.832+0.773.919+0.893.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.590+0.274.601+0.314.5070.303
117score overall qualityeiv_extra:score_overall_quality3.123+0.933.195+1.072.6520.507
118score speech qualityeiv_extra:score_speech_quality1.881+0.501.853+0.311.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.725+0.915.633+0.824.7991.018
120Content usefulness (CU)audiobox:CU6.386+0.746.514+0.865.6491.001
121Production complexity (PC)audiobox:PC1.776-0.341.442-0.552.3311.617
122Production quality (PQ)audiobox:PQ7.637+0.687.497+0.527.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw4.074+0.264.008+0.163.9000.669
124Background quality, rawdnsmos:BAK_raw4.500+1.154.500+1.153.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.884+0.983.834+0.923.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.585+0.193.552+0.133.4880.501
127Background quality, calibrateddnsmos:BAK4.220+0.904.228+0.903.3890.928
128Overall quality, calibrateddnsmos:OVRL3.364+0.863.330+0.812.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.942+1.054.051+1.273.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.025+0.040.005+0.010.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.520-0.770.749-0.360.9470.554
132Agevoiceclap_attribute:Age3.230+0.003.370+0.263.2290.547
133Amusementvoiceclap_attribute:Amusement-0.118-0.890.038-0.620.3920.572
134Angervoiceclap_attribute:Anger0.371-1.230.395-1.180.9920.506
135Arousalvoiceclap_attribute:Arousal0.516-1.450.647-1.241.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.226-0.590.165-0.720.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.842+0.432.864+0.562.7700.167
138Awevoiceclap_attribute:Awe0.511-0.110.529-0.040.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.247-1.220.297-1.020.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.299-1.240.325-1.140.5920.236
141Concentrationvoiceclap_attribute:Concentration1.716+0.941.556+0.641.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant0.912-0.340.808-0.561.0690.468
143Confusionvoiceclap_attribute:Confusion0.629-0.040.605-0.120.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.944+0.400.706-0.180.7800.406
145Contemptvoiceclap_attribute:Contempt0.077-1.860.303-1.200.7090.339
146Contentmentvoiceclap_attribute:Contentment0.625+0.120.610+0.070.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.388-1.280.468-1.000.7600.291
148Disgustvoiceclap_attribute:Disgust0.179-2.150.144-2.320.6250.207
149Distressvoiceclap_attribute:Distress0.415-1.460.773-0.440.9270.350
150Doubtvoiceclap_attribute:Doubt0.359-0.930.329-1.090.5350.188
151Elationvoiceclap_attribute:Elation0.282-0.650.399-0.350.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment-0.001-0.980.118-0.600.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.802+0.520.855+0.720.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.236-0.870.252-0.800.4620.261
155Fearvoiceclap_attribute:Fear0.229-1.830.268-1.650.6180.212
156Gendervoiceclap_attribute:Gender1.489+1.631.380+1.52-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.547-0.410.623-0.130.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.634-0.551.702-0.371.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.383-0.810.415-0.730.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability0.210-1.560.379-1.241.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.266-1.030.378-0.690.6080.332
162Interestvoiceclap_attribute:Interest1.607+0.021.596-0.011.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.006-1.070.097-0.880.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.533+0.040.558+0.140.5230.245
165Longingvoiceclap_attribute:Longing0.364-0.790.432-0.480.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.281-1.600.422-1.190.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive1.472-0.511.569-0.351.7750.594
168Painvoiceclap_attribute:Pain0.128-1.160.062-1.460.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.182-1.070.335-0.560.5030.301
170Pridevoiceclap_attribute:Pride0.641-0.400.625-0.440.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.771+1.402.743+1.312.3130.328
172Reliefvoiceclap_attribute:Relief0.527-1.200.735-0.701.0250.415
173Sadnessvoiceclap_attribute:Sadness0.553-0.630.619-0.420.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.385-0.740.603-0.350.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.235-0.950.255-0.910.6350.419
176Shamevoiceclap_attribute:Shame0.308-0.770.516-0.040.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.613+1.260.458+0.830.1600.360
178Sournessvoiceclap_attribute:Sourness0.328-1.300.513-0.660.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.213-1.140.366-0.720.6230.359
180Teasingvoiceclap_attribute:Teasing-0.062-1.590.116-1.120.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude0.544-1.590.751-1.201.3840.529
182Triumphvoiceclap_attribute:Triumph0.199-1.440.314-1.120.7160.359
183Valencevoiceclap_attribute:Valence0.365+0.590.283+0.46-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached2.028+0.702.098+0.831.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold0.105+0.43-0.119-0.13-0.0660.400
186durationvoiceclap_attribute:duration14.085+0.5013.590+0.4210.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.791+1.103.834+1.183.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.894+1.624.883+1.554.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality3.093+1.113.122+1.182.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.953+0.901.967+1.101.8910.069
191talking speedvoiceclap_attribute:talking_speed8.206-0.229.328+0.089.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
0-0.0125-0.0097-0.0867
10.10980.09160.0659
20.08180.09870.1168
30.0519-0.00460.0193
4-0.0914-0.0529-0.0818
5-0.1597-0.0955-0.1102
6-0.0355-0.0532-0.0789
70.17350.24650.2028
80.00270.0005-0.0523
9-0.0247-0.02820.0234
10-0.0766-0.0494-0.0373
11-0.1604-0.1200-0.1019
12-0.0802-0.0353-0.0876
13-0.0979-0.0858-0.0089
14-0.1111-0.1065-0.0372
150.08320.10200.0548
16-0.0077-0.00530.0999
17-0.0700-0.0816-0.0906
18-0.0565-0.0600-0.0803
19-0.0109-0.0329-0.0094
200.02560.0461-0.0028
210.09010.15100.0782
220.02660.08970.0712
23-0.1177-0.01520.0340
240.15650.06420.1159
25-0.0145-0.03020.0129
26-0.0789-0.0949-0.0817
270.01500.0298-0.0131
28-0.0815-0.0845-0.0959
29-0.01410.0429-0.0727
30-0.0357-0.0363-0.0938
31-0.0316-0.0578-0.0804
32-0.02700.05900.0790
33-0.0631-0.0571-0.0960
34-0.0733-0.0781-0.0781
35-0.00690.0411-0.0704
36-0.1190-0.0803-0.0936
37-0.0500-0.03570.0085
380.10190.05740.1093
39-0.0697-0.0526-0.0878
40-0.02600.04390.0130
410.13080.03020.0243
42-0.0832-0.0803-0.0878
430.14130.08780.1409
44-0.0641-0.0250-0.0833
450.02710.05870.1183
460.02350.05480.0089
470.09200.06070.1460
48-0.0726-0.0858-0.0850
49-0.0641-0.0674-0.0880
50-0.0638-0.0282-0.0333
510.17880.19100.2258
52-0.0186-0.0119-0.0103
530.02330.0038-0.0076
54-0.0657-0.0613-0.0235
550.0503-0.05480.0125
56-0.0799-0.0852-0.0826
57-0.0799-0.08130.0292
580.07260.10260.0955
590.10450.13160.0927
600.02150.08100.0912
610.24720.34070.2036
620.09330.12200.0729
63-0.0648-0.1136-0.0827
64-0.02200.0305-0.0137
65-0.0648-0.0790-0.0710
66-0.0227-0.0832-0.0544
67-0.1288-0.0839-0.0840
68-0.0717-0.0936-0.0920
690.15380.17420.2546
700.0177-0.0252-0.0734
710.01520.03310.0292
72-0.0197-0.01960.0481
730.0127-0.0248-0.0580
740.13540.07190.0865
75-0.0467-0.0658-0.1058
760.04900.01590.1481
77-0.0710-0.0839-0.0795
78-0.0779-0.0749-0.0775
790.12100.11490.1617
80-0.0720-0.0755-0.0789
810.29190.25940.2072
82-0.0519-0.0419-0.0833
830.01220.0419-0.0103
84-0.0365-0.0363-0.0895
850.04210.0845-0.0411
860.00370.02370.0325
87-0.1032-0.1291-0.1005
88-0.0394-0.0248-0.0383
89-0.1012-0.0574-0.0800
90-0.00300.01670.0035
91-0.0575-0.0497-0.0800
92-0.0782-0.0255-0.0606
930.0139-0.01120.0333
940.0237-0.0318-0.0489
95-0.1177-0.1181-0.0492
96-0.1196-0.1342-0.0887
970.0172-0.03440.0319
98-0.0293-0.0610-0.0485
990.14400.12450.1498
1000.14920.09100.0138
101-0.0769-0.0261-0.0842
1020.02700.08390.0153
103-0.0437-0.0826-0.0707
104-0.0427-0.0175-0.0851
105-0.00620.01360.0866
1060.0641-0.01100.0188
107-0.0763-0.0681-0.0811
108-0.0887-0.0845-0.0945
1090.1045-0.02210.0540
110-0.1749-0.1226-0.0679
1110.03860.00340.1698
112-0.0999-0.1258-0.0858
113-0.0316-0.03500.0229
1140.13410.18450.1251
1150.11240.14200.1006
1160.14130.10390.1245
1170.08380.05000.0426
118-0.0815-0.0661-0.0848
119-0.00310.0190-0.0612
1200.06340.15030.0625
1210.15970.12840.1697
122-0.0999-0.1039-0.0895
1230.05160.00980.0455
1240.03940.0003-0.0252
125-0.1058-0.1374-0.0868
1260.02940.01310.0556
127-0.1131-0.1007-0.0547
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
00.15020.09730.1238
1-0.0154-0.0511-0.0102
2-0.0779-0.0419-0.0559
30.05070.02650.0541
40.02230.0007-0.0414
50.0066-0.0020-0.0475
60.03200.0043-0.0139
70.02500.06980.0335
80.0261-0.0233-0.0512
90.08050.08030.0709
100.0284-0.01610.0608
110.0034-0.0424-0.0615
120.08300.03810.0808
130.19260.13820.1431
14-0.0315-0.0531-0.0649
150.10520.02600.0215
160.04910.06640.0252
17-0.0231-0.0565-0.0243
180.12170.14570.1069
19-0.1736-0.1416-0.0721
20-0.0792-0.0698-0.0455
21-0.0353-0.0400-0.0422
22-0.0033-0.0117-0.0310
23-0.0982-0.1021-0.0440
24-0.0291-0.0272-0.0388
25-0.0200-0.0073-0.0126
26-0.03340.0311-0.0032
27-0.0453-0.0575-0.0370
28-0.03120.00190.0266
290.0402-0.0032-0.0048
30-0.0596-0.0813-0.0396
310.18120.15720.2399
32-0.0261-0.0245-0.0043
330.15590.13340.0756
340.09890.11710.1420
35-0.0244-0.0225-0.0586
360.02440.0259-0.0203
370.0277-0.01770.0262
380.0147-0.01080.0712
39-0.0874-0.0892-0.0384
40-0.0520-0.0257-0.0281
41-0.0355-0.0619-0.0522
420.0383-0.0163-0.0219
430.06270.11230.0936
44-0.0058-0.0277-0.0547
45-0.0105-0.00870.0088
460.01850.0265-0.0134
47-0.0114-0.0390-0.0001
48-0.0261-0.0568-0.0485
49-0.0634-0.0820-0.0809
50-0.0230-0.0300-0.0321
510.0074-0.02210.0737
520.06080.05580.0735
53-0.0371-0.0476-0.0440
540.12100.15720.1562
55-0.0053-0.01310.0048
560.04010.00550.0475
570.12800.15040.1825
580.0000-0.0260-0.0118
590.09820.06190.0817
600.00490.0076-0.0345
610.0139-0.0180-0.0422
62-0.0298-0.0281-0.0399
63-0.00400.0094-0.0494
64-0.0322-0.1103-0.0506
65-0.01870.0362-0.0497
660.0624-0.0103-0.0020
670.02530.00760.0778
680.03040.01110.0142
690.06270.0120-0.0075
70-0.0523-0.0456-0.0659
71-0.0776-0.0694-0.0476
720.02680.01000.1437
73-0.01840.09460.0485
74-0.1096-0.1014-0.0491
750.01270.04340.0947
76-0.05200.00400.0541
77-0.0352-0.0109-0.0320
78-0.0415-0.0417-0.0663
79-0.0162-0.0572-0.0486
80-0.0497-0.0919-0.0483
81-0.0602-0.0718-0.0544
820.07350.0305-0.0172
83-0.0050-0.0257-0.0290
840.02910.02120.0388
85-0.0356-0.0708-0.0486
86-0.0817-0.0572-0.0433
87-0.0394-0.0199-0.0198
880.19900.27770.2881
89-0.0409-0.0414-0.0439
900.08810.11300.0353
910.10580.19470.1643
92-0.0166-0.0134-0.0077
930.06020.07280.1024
940.0057-0.0541-0.0109
95-0.0168-0.0514-0.0507
96-0.0024-0.0334-0.0434
97-0.0196-0.0732-0.0163
980.11660.08580.0749
990.01810.01400.0028
1000.09570.03970.0317
101-0.0887-0.0677-0.0459
1020.09820.06980.0994
1030.0208-0.00680.0343
104-0.0012-0.0776-0.0527
105-0.0779-0.1028-0.0632
106-0.0387-0.0701-0.0374
107-0.0339-0.0121-0.0435
1080.01800.02720.0253
109-0.0111-0.0081-0.0280
110-0.0586-0.0824-0.0578
1110.00530.0084-0.0400
1120.07760.02840.0806
113-0.0539-0.0204-0.0828
1140.05990.01040.0474
1150.05010.08340.1147
1160.0101-0.0402-0.0365
1170.02800.04460.0662
118-0.0779-0.0422-0.0634
1190.07790.11910.0633
1200.01090.03280.0438
121-0.0843-0.0800-0.0522
122-0.0643-0.0589-0.0458
1230.0106-0.0262-0.0501
1240.0135-0.00760.0155
1250.07190.10480.0798
126-0.0173-0.0207-0.0110
127-0.0249-0.0691-0.0499
128-0.0093-0.0045-0.0148
1290.09950.03300.0814
130-0.0688-0.0619-0.0481
131-0.0463-0.0009-0.0425
1320.01940.04630.0374
133-0.0211-0.0371-0.0480
1340.06910.0376-0.0116
1350.0198-0.0238-0.0167
136-0.0246-0.0606-0.0520
1370.0239-0.0035-0.0451
138-0.0602-0.0538-0.0428
1390.01770.04150.0485
1400.0323-0.0613-0.0257
1410.00600.0228-0.0233
1420.0000-0.0193-0.0207
143-0.0138-0.0892-0.0727
1440.02880.02400.0036
145-0.0970-0.1205-0.0617
146-0.0428-0.08300.0187
147-0.0548-0.0551-0.0431
148-0.0624-0.0165-0.0535
149-0.0200-0.0294-0.0132
150-0.0398-0.0294-0.0478
151-0.0817-0.0790-0.0553
152-0.0299-0.0514-0.0538
1530.0387-0.01340.0079
154-0.01830.0279-0.0046
1550.09630.08850.0446
1560.03900.07280.0212
1570.11340.09190.1351
158-0.0722-0.0335-0.0605
159-0.0849-0.0391-0.0568
1600.03720.04870.0330
1610.09760.03370.0192
162-0.0352-0.0939-0.0563
163-0.0326-0.0237-0.0457
164-0.0596-0.0483-0.0426
1650.07190.10890.1147
166-0.0577-0.1144-0.0565
167-0.01350.0080-0.0615
168-0.0231-0.00880.0059
1690.07350.02300.0725
1700.0041-0.01530.0118
171-0.02250.01040.0105
172-0.1369-0.0980-0.0562
173-0.0339-0.0378-0.0569
174-0.0054-0.0470-0.0446
1750.03010.05550.0332
176-0.0608-0.0345-0.0413
177-0.0173-0.0262-0.0147
1780.10520.05620.0521
179-0.0688-0.0667-0.0598
1800.0005-0.0174-0.0507
1810.10330.08000.1383
182-0.0554-0.0243-0.0419
1830.03800.02910.0617
184-0.0006-0.0470-0.0446
1850.08740.05410.0441
186-0.0577-0.0953-0.0585
1870.12290.08200.0831
1880.0115-0.0218-0.0312
1890.03850.05210.0729
190-0.1103-0.0960-0.0842
191-0.0371-0.0140-0.0284
192-0.0551-0.00480.0237
193-0.0230-0.0623-0.0511
1940.0309-0.00570.0353
195-0.0450-0.0397-0.0355
196-0.0261-0.0289-0.0472
197-0.0132-0.0335-0.0679
198-0.0716-0.0565-0.0469
199-0.0672-0.0688-0.0533
2000.0147-0.01790.0165
2010.07640.0305-0.0118
2020.02530.0154-0.0129
2030.03580.0449-0.0156
2040.09570.05750.0670
2050.0379-0.03930.0062
2060.0504-0.00400.0029
2070.01110.0303-0.0027
208-0.0485-0.0541-0.0503
2090.03250.01130.0432
210-0.0336-0.0254-0.0277
2110.02170.04660.0964
212-0.1058-0.0878-0.0710
2130.01200.0442-0.0070
214-0.0325-0.0088-0.0201
215-0.0539-0.0259-0.0408
2160.07920.07660.0346
2170.18000.09800.1160
2180.11150.07050.1049
219-0.0127-0.0664-0.0457
220-0.1147-0.0953-0.0536
2210.05070.07620.1191
2220.0665-0.0140-0.0122
2230.12670.11160.1215
224-0.0425-0.0473-0.0443
225-0.0577-0.0335-0.0404
2260.09820.05110.0640
2270.04910.09120.0807
228-0.0586-0.0800-0.0786
2290.03610.02790.0399
230-0.0055-0.00310.0151
231-0.0513-0.0616-0.0546
2320.09250.12870.1215
233-0.02040.0190-0.0267
2340.0063-0.0378-0.0412
2350.09130.03570.0777
236-0.0453-0.0504-0.0417
2370.02870.03710.0263
238-0.0001-0.0168-0.0201
239-0.0672-0.1041-0.0502
240-0.0713-0.0885-0.0495
2410.0415-0.03110.0626
242-0.0862-0.0871-0.0715
243-0.00730.00150.0343
244-0.0260-0.0218-0.0516
2450.15210.10760.1654
246-0.0085-0.0191-0.0393
247-0.0202-0.0504-0.0409
248-0.0459-0.0677-0.0289
249-0.0007-0.02070.0006
Source: Emolia / Emilia source archive · SHA-256 0ac4c02fa170fc9e…
p01Speech + burstsP3 curated burst test

Amusement + 5 known burst(s)

Synthetic speech with inserted bursts · P3 synthetic speech/burst mixture · 5.0 s

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Intoxication Altered States of Consciousness
z +4.71 / +4.19raw 2.61 / 2.32
Embarrassment
z +3.97 / +4.03raw 1.75 / 1.77
Amusement
z +3.57 / +3.57raw 3.16 / 3.16
Teasing
z +2.44 / +2.68raw 1.33 / 1.46
Pleasure Ecstasy
z +1.83 / +2.13raw 1.14 / 1.32
Elation
z +1.96 / +1.95raw 1.44 / 1.43
Confusion
z +1.46 / +1.68raw 0.79 / 0.91

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Volatility (VOLT)
z +2.55 / +2.44raw 4.84 / 4.75
Content appropriateness (EXPL)
z +2.46 / +2.35raw 0.88 / 0.84
Playful style (S_PLAY)
z +1.91 / +2.01raw 4.86 / 4.99
Casual style (S_CASU)
z +1.96 / +1.81raw 6.29 / 6.10
Roughness (ROUG)
z +1.70 / +1.52raw 3.55 / 3.38
Disfluency (DFLU)
z +1.69 / +1.48raw 4.49 / 4.25
Dynamic arc (DARC)
z +1.50 / +1.62raw 3.69 / 3.85

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z +1.30 / +1.38raw 4.44 / 4.57
Vocal-burst blend (0–10)z -2.22 / -2.21raw 0.59 / 0.63
Content enjoyment (CE)z -0.69 / -0.76raw 4.09 / 4.03
Production quality (PQ)z -1.24 / -1.80raw 5.95 / 5.46
Speech signal quality, calibratedz -1.83 / -2.42raw 2.57 / 2.28
Background quality, calibratedz -1.44 / -1.29raw 2.05 / 2.19
Overall quality, calibratedz -1.68 / -1.70raw 1.82 / 1.81
P.808 MOS estimatez -1.25 / -1.27raw 2.80 / 2.79
score content enjoymentz -1.73 / -1.98raw 3.98 / 3.91
score overall qualityz -1.50 / -1.48raw 1.89 / 1.90
Background Noisez +0.81 / +0.33raw 1.02 / 0.89
Tempo (TEMP)z +0.57 / +0.62raw 2.80 / 2.85
Arousal (AROU)z +0.76 / +0.79raw 3.23 / 3.26
Valence (VALN)z +0.15 / +0.32raw 2.50 / 2.68
Recording quality (RCQL)z -0.74 / -0.96raw 2.09 / 1.90
Background noise (BKGN)z -0.71 / -0.81raw 2.16 / 2.09
talking speedz -0.59 / -0.08raw 6.83 / 8.74
Characters per second (CPS): 4.64 raw · z -1.17

P3 metadata teacher estimate; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.4400.377
Identity · 250D0.4490.368

Vocal-burst regions

Known
Tiny
Base
0 s5.0 s

Purple marks known inserted burst intervals; other vocal sounds may be unlabeled.

Tiny: 1 regions, count-head estimate 3.10; Base: 1 regions, count-head estimate 2.72.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
Tiny10.184.624.44
Base10.184.624.44

Known inserted intervals: 4.03–4.36 s, 0.17–0.72 s, 3.18–3.83 s, 0.26–4.18 s, 4.12–4.60 s

All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.061+0.090.184+0.310.0070.570
1Amusementemo_Amusement3.162+3.573.162+3.57-0.0010.885
2Angeremo_Anger0.000-0.01-0.086-0.150.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.217+0.370.167+0.280.0050.572
4Aweemo_Awe-0.029-0.07-0.059-0.140.0000.427
5Bitternessemo_Bitterness-0.065-0.19-0.021-0.080.0130.415
6Concentrationemo_Concentration-0.066-0.940.011-0.840.6340.741
7Confusionemo_Confusion0.789+1.460.909+1.680.0010.540
8Contemplationemo_Contemplation0.032-0.080.018-0.110.0820.598
9Contemptemo_Contempt-0.075-0.160.010+0.02-0.0000.464
10Contentmentemo_Contentment0.528+1.030.651+1.270.0000.514
11Disappointmentemo_Disappointment-0.028-0.04-0.014-0.020.0000.700
12Disgustemo_Disgust0.039+0.06-0.017-0.050.0070.489
13Distressemo_Distress0.053+0.070.009+0.010.0000.821
14Doubtemo_Doubt0.290+0.610.274+0.580.0060.466
15Elationemo_Elation1.440+1.961.434+1.950.0120.728
16Embarrassmentemo_Embarrassment1.747+3.971.774+4.030.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.075+0.080.045+0.040.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.301+0.510.225+0.370.0280.539
19Fearemo_Fear-0.030-0.07-0.001-0.030.0150.610
20Helplessnessemo_Helplessness-0.031-0.040.033+0.05-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.543+0.560.525+0.540.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.156+0.010.143-0.010.1490.705
23Infatuationemo_Infatuation0.026+0.080.193+0.39-0.0150.533
24Interestemo_Interest0.871-0.790.939-0.721.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness2.606+4.712.319+4.19-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy-0.020-0.180.014-0.100.0590.442
27Longingemo_Longing-0.079-0.260.058-0.050.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.074+0.040.050-0.010.0550.534
29Painemo_Pain0.142+0.250.101+0.18-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy1.144+1.831.325+2.130.0170.614
31Prideemo_Pride-0.033-0.07-0.029-0.060.0020.509
32Reliefemo_Relief-0.012-0.070.010-0.030.0330.681
33Sadnessemo_Sadness-0.007-0.01-0.028-0.040.0000.759
34Sexual Lustemo_Sexual_Lust0.685+1.330.756+1.470.0300.494
35Shameemo_Shame0.068+0.110.087+0.150.0130.510
36Sournessemo_Sourness-0.013-0.02-0.015-0.02-0.0050.439
37Teasingemo_Teasing1.329+2.441.460+2.680.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.037-0.010.030-0.020.0410.697
39Triumphemo_Triumph-0.043-0.07-0.038-0.06-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg2.093-0.721.643-1.082.9841.238
41Arousal (AROU)vn_AROU_reg3.229+0.763.264+0.792.3591.147
42Arousal shift (ARSH)vn_ARSH_reg2.180-0.462.013-0.652.5800.867
43Vocal attack (ATCK)vn_ATCK_reg2.504+0.262.741+0.452.1621.295
44Background noise (BKGN)vn_BKGN_reg2.164-0.712.090-0.812.7030.761
45Brightness (BRGT)vn_BRGT_reg1.223-0.951.201-0.971.9910.813
46Speech chunking (CHNK)vn_CHNK_reg0.859-0.800.796-0.851.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg0.588-1.510.528-1.582.0180.945
48Cognitive load (COGL)vn_COGL_reg0.630-1.380.037-1.942.0941.061
49Dynamic arc (DARC)vn_DARC_reg3.688+1.503.847+1.621.6151.380
50Disfluency (DFLU)vn_DFLU_reg4.495+1.694.245+1.482.4881.184
51Emphasis (EMPH)vn_EMPH_reg0.877-1.140.472-1.402.6071.524
52Esthetics (ESTH)vn_ESTH_reg1.479-1.181.431-1.242.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.879+2.460.843+2.350.0860.322
54Focus (FOCS)vn_FOCS_reg0.958-1.260.663-1.532.3441.102
55Fullness (FULL)vn_FULL_reg1.861+0.091.554-0.251.7780.889
56Perceived gender (GEND)vn_GEND_reg4.590+0.344.572+0.334.0551.588
57Harmonicity (HARM)vn_HARM_reg3.338+1.033.345+1.032.4800.836
58Metallic character (METL)vn_METL_reg1.570-0.051.748+0.131.6210.958
59Pitch range (RANG)vn_RANG_reg3.129+0.243.233+0.362.9140.897
60Recording quality (RCQL)vn_RCQL_reg2.094-0.741.897-0.962.7540.888
61Vocal register (REGS)vn_REGS_reg1.707+0.401.788+0.461.1981.268
62Respiration (RESP)vn_RESP_reg4.041+1.294.225+1.462.6801.056
63Roughness (ROUG)vn_ROUG_reg3.545+1.703.380+1.522.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg1.603-0.631.720-0.492.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg2.165+0.022.074-0.092.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg0.712-1.190.644-1.261.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg0.929-0.600.657-0.931.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg3.031+0.232.861+0.072.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg1.981-0.272.032-0.202.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg1.944-0.161.728-0.432.0700.803
71Smoothness (SMTH)vn_SMTH_reg1.177-0.880.817-1.212.1431.093
72Stance (STNC)vn_STNC_reg1.818-0.721.724-0.802.6331.137
73Speech structure (STRU)vn_STRU_reg0.111-2.110.011-2.222.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg0.916-0.320.585-0.561.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg-0.266-1.30-0.306-1.341.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg2.894+0.953.017+1.041.5601.401
77Casual style (S_CASU)vn_S_CASU_reg6.292+1.966.103+1.813.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg4.308+1.304.145+1.182.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg3.661+0.823.571+0.742.7211.148
80Formal style (S_FORM)vn_S_FORM_reg-0.204-1.37-0.117-1.291.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg0.271-1.990.062-2.152.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg-0.330-1.13-0.162-1.021.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.031-0.66-0.051-0.810.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg4.858+1.914.990+2.012.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg2.112+0.702.222+0.781.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg1.405-0.790.997-1.092.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg0.451-0.160.110-0.440.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg1.130-0.290.905-0.431.5971.618
89Tempo (TEMP)vn_TEMP_reg2.795+0.572.851+0.622.1171.190
90Tension (TENS)vn_TENS_reg2.518+1.132.316+0.891.5590.852
91Valence (VALN)vn_VALN_reg2.499+0.152.680+0.322.3281.118
92Valence shift (VALS)vn_VALS_reg3.485+0.903.304+0.732.5181.079
93Velocity flux (VFLX)vn_VFLX_reg3.517+1.043.489+1.022.2561.208
94Volatility (VOLT)vn_VOLT_reg4.842+2.554.747+2.442.5860.886
95Vulnerability (VULN)vn_VULN_reg4.172+1.553.878+1.252.6480.984
96Warmth (WARM)vn_WARM_reg1.545-1.141.538-1.142.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_64.445+1.304.570+1.382.3321.621
98Vocal-burst blend (0–10)blend_0_100.592-2.220.635-2.216.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.830-0.780.851-0.721.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age2.635-2.102.699-1.883.2500.293
101Arousaleiv_extra:Arousal2.237+2.182.158+2.001.2890.435
102Authenticityeiv_extra:Authenticity2.725+0.382.731+0.422.6670.153
103Background Noiseeiv_extra:Background_Noise1.020+0.810.894+0.330.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.910+2.871.933+2.951.0400.303
105Gendereiv_extra:Gender-0.196+0.42-0.179+0.44-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.499-1.311.530-1.191.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive2.469+1.302.495+1.351.8590.470
108Recording Qualityeiv_extra:Recording_Quality1.408-2.221.434-2.142.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous2.537+4.722.441+4.440.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh1.027+2.331.067+2.460.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant-0.212-1.81-0.215-1.820.3760.324
112Valenceeiv_extra:Valence1.252+3.281.508+3.950.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached0.921-1.810.911-1.841.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold0.946+2.981.002+3.170.0510.300
115score background qualityeiv_extra:score_background_quality2.321-1.422.310-1.433.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment3.981-1.733.908-1.984.5070.303
117score overall qualityeiv_extra:score_overall_quality1.890-1.501.902-1.482.6520.507
118score speech qualityeiv_extra:score_speech_quality1.766-0.281.791-0.111.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE4.093-0.694.025-0.764.7991.018
120Content usefulness (CU)audiobox:CU4.907-0.744.435-1.215.6491.001
121Production complexity (PC)audiobox:PC3.237+0.563.180+0.532.3311.617
122Production quality (PQ)audiobox:PQ5.955-1.245.460-1.807.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw2.621-1.912.239-2.483.9000.669
124Background quality, rawdnsmos:BAK_raw1.837-1.222.004-1.073.2111.124
125Overall quality, rawdnsmos:OVRL_raw1.827-1.551.814-1.563.0850.812
126Speech signal quality, calibrateddnsmos:SIG2.573-1.832.275-2.423.4880.501
127Background quality, calibrateddnsmos:BAK2.051-1.442.189-1.293.3890.928
128Overall quality, calibrateddnsmos:OVRL1.823-1.681.809-1.702.8400.606
129P.808 MOS estimatednsmos:P808_MOS2.802-1.252.790-1.273.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p1.411+2.181.315+2.030.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.747-0.360.886-0.110.9470.554
132Agevoiceclap_attribute:Age2.766-0.852.646-1.073.2290.547
133Amusementvoiceclap_attribute:Amusement3.347+5.173.365+5.200.3920.572
134Angervoiceclap_attribute:Anger0.296-1.380.105-1.750.9920.506
135Arousalvoiceclap_attribute:Arousal2.364+1.522.437+1.631.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.928+0.870.832+0.680.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.759-0.072.725-0.272.7700.167
138Awevoiceclap_attribute:Awe-0.089-2.42-0.130-2.580.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.784+0.820.726+0.600.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.202-1.660.173-1.780.5920.236
141Concentrationvoiceclap_attribute:Concentration0.083-2.13-0.083-2.441.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.679+1.301.731+1.411.0690.468
143Confusionvoiceclap_attribute:Confusion0.622-0.070.651+0.020.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.116-1.640.085-1.710.7800.406
145Contemptvoiceclap_attribute:Contempt0.369-1.000.369-1.000.7090.339
146Contentmentvoiceclap_attribute:Contentment0.554-0.110.628+0.130.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.207-1.900.179-1.990.7600.291
148Disgustvoiceclap_attribute:Disgust0.382-1.170.283-1.650.6250.207
149Distressvoiceclap_attribute:Distress0.084-2.410.073-2.440.9270.350
150Doubtvoiceclap_attribute:Doubt0.406-0.680.402-0.710.5350.188
151Elationvoiceclap_attribute:Elation1.336+2.011.352+2.050.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment1.530+3.891.525+3.880.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.009-2.46-0.008-2.530.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.428-0.130.397-0.250.4620.261
155Fearvoiceclap_attribute:Fear0.307-1.460.298-1.510.6180.212
156Gendervoiceclap_attribute:Gender0.001+0.150.022+0.17-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.301-1.330.265-1.460.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.402-1.171.420-1.121.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.761+0.140.748+0.110.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability0.286-1.420.290-1.411.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.407-0.610.457-0.460.6080.332
162Interestvoiceclap_attribute:Interest1.326-0.621.259-0.771.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness1.633+2.441.626+2.430.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.072-1.840.137-1.580.5230.245
165Longingvoiceclap_attribute:Longing0.107-1.960.220-1.440.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.600-0.670.541-0.840.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive2.767+1.672.925+1.941.7750.594
168Painvoiceclap_attribute:Pain0.167-0.980.150-1.060.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy1.354+2.831.335+2.760.5030.301
170Pridevoiceclap_attribute:Pride0.051-2.09-0.042-2.350.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality1.779-1.631.682-1.932.3130.328
172Reliefvoiceclap_attribute:Relief0.403-1.500.478-1.321.0250.415
173Sadnessvoiceclap_attribute:Sadness0.335-1.330.325-1.360.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous3.290+4.383.504+4.750.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.745+0.260.678+0.100.6350.419
176Shamevoiceclap_attribute:Shame0.322-0.720.344-0.650.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.897+2.051.066+2.520.1600.360
178Sournessvoiceclap_attribute:Sourness0.213-1.700.203-1.730.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.268-0.990.275-0.970.6230.359
180Teasingvoiceclap_attribute:Teasing1.586+2.761.604+2.810.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude0.354-1.950.411-1.841.3840.529
182Triumphvoiceclap_attribute:Triumph0.157-1.560.168-1.530.7160.359
183Valencevoiceclap_attribute:Valence2.133+3.372.302+3.64-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached1.084-1.000.841-1.431.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold1.141+3.021.229+3.24-0.0660.400
186durationvoiceclap_attribute:duration5.648-0.834.757-0.9710.9146.354
187score background qualityvoiceclap_attribute:score_background_quality2.619-1.262.646-1.203.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.572-0.574.507-1.014.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.104-1.292.088-1.332.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.937+0.671.959+0.981.8910.069
191talking speedvoiceclap_attribute:talking_speed6.829-0.598.741-0.089.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.00090.0484-0.0696
1-0.0053-0.1062-0.0690
20.07480.0316-0.0135
3-0.00720.0291-0.0367
4-0.0252-0.0106-0.0657
50.02670.03070.0492
6-0.1183-0.0704-0.0634
7-0.0254-0.0549-0.0131
8-0.0492-0.0715-0.0423
9-0.0733-0.0664-0.0095
100.03510.0841-0.0650
11-0.0931-0.0524-0.0817
12-0.0437-0.05710.0515
13-0.0767-0.0639-0.0154
140.08010.0029-0.0474
150.0485-0.03830.1118
16-0.0652-0.06500.0479
170.03050.0487-0.0079
18-0.0649-0.0794-0.0645
190.18540.14300.2097
20-0.0527-0.0433-0.0635
21-0.0060-0.0747-0.0313
220.14650.11480.0865
230.09730.18630.0331
24-0.0397-0.02350.0181
25-0.0549-0.04660.0157
26-0.1084-0.0968-0.0656
27-0.1068-0.09390.1078
280.01130.01130.1198
290.08620.07190.0958
30-0.0893-0.1163-0.0753
310.0706-0.0283-0.0645
320.04790.0899-0.0587
33-0.1106-0.1437-0.0770
34-0.01720.0502-0.0309
35-0.0580-0.05670.0195
36-0.0018-0.02980.0481
37-0.0828-0.1372-0.0828
38-0.0641-0.10690.1194
39-0.01250.01810.0890
400.0098-0.0084-0.0564
41-0.0286-0.1098-0.0244
420.09770.0495-0.0224
43-0.0908-0.12780.0823
44-0.0843-0.1148-0.0671
450.21520.24270.2278
46-0.0900-0.1127-0.0659
470.07020.12780.2023
48-0.0137-0.0578-0.0682
490.20300.2340-0.0171
50-0.0022-0.0339-0.0715
510.0414-0.0165-0.0400
520.0155-0.0235-0.0316
53-0.0408-0.05850.0479
540.1122-0.0211-0.0591
55-0.1206-0.1199-0.0673
560.0126-0.0469-0.0663
57-0.0858-0.1286-0.0385
580.07550.05160.0750
59-0.0320-0.10540.0560
600.19690.1993-0.0154
610.11680.04980.1918
620.26860.27010.1799
63-0.0286-0.0477-0.0664
64-0.1160-0.0509-0.0709
65-0.0202-0.03120.0256
660.07360.03480.0433
670.08780.1639-0.0674
68-0.0855-0.0856-0.0739
690.0486-0.02030.0656
70-0.1450-0.1062-0.0696
71-0.1305-0.1387-0.0775
720.09190.10910.0264
73-0.1251-0.1119-0.0278
74-0.0099-0.0672-0.0098
75-0.0683-0.0805-0.0849
76-0.0370-0.06790.0136
77-0.0721-0.0432-0.0638
78-0.0399-0.0312-0.0622
79-0.0265-0.02760.1785
800.0179-0.0422-0.0633
81-0.0138-0.03320.0592
82-0.02880.0093-0.0585
83-0.0025-0.0127-0.0715
84-0.0805-0.0334-0.0314
85-0.0515-0.0545-0.0338
860.05230.01400.0515
87-0.1473-0.1257-0.0807
880.10840.0056-0.0649
890.03190.0180-0.0641
900.02500.02830.0132
91-0.0950-0.1228-0.0643
920.21980.18780.0580
930.0725-0.01190.0141
94-0.08240.0100-0.0323
950.0254-0.0348-0.0242
96-0.0870-0.0273-0.0964
970.12210.1560-0.0539
98-0.0977-0.08630.0419
990.0660-0.06170.0988
100-0.0206-0.0773-0.0475
101-0.1435-0.1517-0.0235
102-0.0265-0.0218-0.0466
1030.11980.08090.0876
104-0.0507-0.0361-0.0310
1050.0034-0.00650.2920
106-0.1450-0.0697-0.0840
1070.0238-0.0585-0.0651
1080.04410.0863-0.0758
109-0.0633-0.0221-0.0685
1100.03220.0444-0.1039
111-0.0729-0.0877-0.0811
112-0.03010.05020.0282
113-0.0752-0.05200.0845
114-0.0683-0.06170.1329
1150.02350.04600.0678
1160.0633-0.01490.2361
117-0.1519-0.04370.0269
118-0.0931-0.0899-0.0681
119-0.1229-0.14080.0197
1200.23350.13510.2057
1210.11980.13790.3176
122-0.02140.0133-0.0718
123-0.02630.0274-0.0735
124-0.0591-0.0823-0.0975
125-0.1007-0.1040-0.0697
126-0.0588-0.04980.1172
127-0.0523-0.0560-0.0742
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.0087-0.0057-0.0012
1-0.0643-0.0705-0.0417
2-0.0353-0.04350.0270
30.06720.05730.0116
4-0.0433-0.0712-0.0392
5-0.0239-0.0431-0.0450
6-0.0372-0.02230.0118
70.10730.06620.0488
80.00840.0630-0.0162
90.02030.05940.0313
10-0.00850.03110.0790
11-0.0690-0.0644-0.0583
12-0.0690-0.0512-0.0369
130.01190.0152-0.0168
14-0.0316-0.00630.1003
15-0.0465-0.0379-0.0508
160.22730.19610.0683
17-0.1131-0.1131-0.0517
180.00590.01950.0594
190.00280.03690.0445
20-0.0828-0.0551-0.0431
21-0.0749-0.06120.0576
22-0.0619-0.0523-0.0396
23-0.00340.0426-0.0417
24-0.0743-0.0594-0.0368
25-0.02200.03100.0966
26-0.0065-0.0310-0.0286
27-0.0728-0.0838-0.0350
28-0.0733-0.07950.0726
29-0.0648-0.0702-0.0064
300.02720.0187-0.0374
310.21240.09660.2228
320.05710.1031-0.0443
33-0.0007-0.04620.0012
340.07380.0290-0.0432
35-0.0656-0.0798-0.0215
36-0.0629-0.0612-0.0455
37-0.0414-0.0734-0.0298
380.0425-0.01630.0136
39-0.0443-0.0648-0.0009
40-0.0961-0.0995-0.0499
41-0.0219-0.0285-0.0495
42-0.0315-0.0376-0.0428
430.01580.0002-0.0509
44-0.0446-0.0211-0.0518
45-0.0435-0.0551-0.0062
46-0.0770-0.0766-0.0633
470.06480.1024-0.0564
48-0.0489-0.0156-0.0180
49-0.0882-0.0741-0.0570
500.09240.09060.0494
510.01240.0530-0.0382
52-0.0797-0.0898-0.0405
53-0.0470-0.0687-0.0417
540.06240.05080.1302
550.04750.03380.0362
56-0.0343-0.03270.0213
57-0.0358-0.0428-0.0078
58-0.0204-0.0501-0.0413
59-0.0409-0.0727-0.0213
600.00420.00890.1555
610.03320.1238-0.0400
62-0.01020.0009-0.0378
63-0.00140.0140-0.0468
64-0.0966-0.0938-0.0479
65-0.0183-0.0129-0.0526
660.02380.02150.0487
670.03250.0016-0.0453
68-0.0523-0.0216-0.0464
690.0656-0.01480.1156
70-0.0844-0.09310.0664
71-0.0876-0.0759-0.0451
72-0.0181-0.0199-0.0477
730.0107-0.03110.0157
74-0.0568-0.0734-0.0184
75-0.0528-0.06260.0016
760.0011-0.0399-0.0106
77-0.0109-0.0194-0.0512
780.03690.0288-0.0117
79-0.0754-0.0523-0.0461
800.04910.0608-0.0457
81-0.0765-0.0487-0.0516
820.0093-0.02050.0627
830.0099-0.0020-0.0382
84-0.0323-0.00460.0326
85-0.0196-0.03540.0702
86-0.0940-0.0659-0.0410
87-0.0137-0.0630-0.0039
880.18590.15680.1382
89-0.0129-0.04240.1008
90-0.0116-0.0109-0.0456
910.11260.10240.0351
920.04650.02650.0776
930.00200.0085-0.0522
94-0.1083-0.1052-0.0048
950.0068-0.0209-0.0480
96-0.0526-0.0175-0.0411
970.03200.0458-0.0259
980.06160.01090.0123
99-0.0012-0.0143-0.0576
1000.15400.12530.1256
1010.0085-0.0004-0.0435
1020.11360.1095-0.0514
1030.04650.03360.0609
104-0.0696-0.0780-0.0599
105-0.1041-0.0988-0.0598
106-0.0388-0.0189-0.0354
107-0.0264-0.0202-0.0412
108-0.0443-0.0494-0.0342
109-0.0802-0.0909-0.0442
110-0.0451-0.0898-0.0547
111-0.0640-0.1009-0.0379
112-0.0966-0.0755-0.0080
113-0.0008-0.0203-0.0784
114-0.0560-0.0795-0.0010
1150.14020.07090.1644
116-0.00680.0179-0.0825
1170.04350.01600.0240
1180.0177-0.0038-0.0174
1190.04140.08590.0567
120-0.0475-0.04980.1506
121-0.0754-0.0719-0.0195
122-0.0222-0.04010.0236
1230.05360.07020.1274
124-0.0539-0.0421-0.0459
125-0.0446-0.0442-0.0447
1260.0316-0.0109-0.0884
127-0.0690-0.0326-0.0473
128-0.0754-0.06300.1912
129-0.0042-0.01050.0903
130-0.0552-0.0616-0.0462
131-0.0422-0.0379-0.0403
132-0.0494-0.0103-0.0487
133-0.0366-0.04620.1251
1340.01140.0112-0.0517
1350.03610.0311-0.0463
136-0.1009-0.0848-0.0493
137-0.0775-0.0838-0.0427
138-0.0797-0.0684-0.0406
139-0.0523-0.06550.1073
1400.0264-0.01310.0309
141-0.0441-0.0841-0.0455
142-0.0555-0.0444-0.0197
143-0.0759-0.0644-0.0861
144-0.0645-0.0787-0.0479
145-0.0993-0.07590.0152
1460.0168-0.0062-0.0551
1470.03040.01850.0511
148-0.0945-0.10240.0678
149-0.0094-0.03060.1002
150-0.0018-0.02250.0114
151-0.0632-0.0576-0.0524
152-0.01040.0231-0.0510
153-0.0531-0.0798-0.0362
1540.04430.06340.1158
155-0.0850-0.1045-0.0600
156-0.0195-0.05010.1265
1570.0173-0.04080.0367
158-0.02200.00590.0311
1590.03530.0841-0.0427
160-0.0328-0.0455-0.0437
161-0.0706-0.03950.1062
162-0.0871-0.0626-0.0534
163-0.1004-0.1131-0.0512
164-0.0579-0.0413-0.0404
1650.09610.06620.0934
166-0.0119-0.0208-0.0535
167-0.0621-0.0533-0.0583
1680.04140.02200.0643
1690.01900.0141-0.0525
170-0.0701-0.0422-0.0430
1710.05120.0392-0.0405
1720.00890.0071-0.0533
173-0.0690-0.04400.0109
174-0.0366-0.05910.0132
175-0.0082-0.0150-0.0180
176-0.0412-0.0365-0.0391
177-0.0550-0.0616-0.0223
1780.10940.07660.0960
179-0.0552-0.0651-0.0505
180-0.0309-0.0834-0.0526
1810.07910.10020.1079
182-0.02270.0064-0.0039
1830.12590.13600.0806
184-0.0770-0.0469-0.0467
185-0.0050-0.0252-0.0550
186-0.0621-0.0583-0.0554
1870.02690.0180-0.0523
188-0.02340.0069-0.0432
1890.00760.0358-0.0237
1900.0039-0.0177-0.0622
191-0.0267-0.04370.1111
192-0.1237-0.1417-0.0557
1930.11100.07910.1255
194-0.0462-0.04690.0864
1950.06800.04900.0599
196-0.0860-0.0601-0.0420
1970.09820.09160.1381
198-0.0834-0.0684-0.0444
1990.05440.1102-0.0392
200-0.0149-0.0433-0.0531
2010.04200.0042-0.0433
202-0.0544-0.0626-0.0095
203-0.0406-0.0056-0.0573
204-0.0441-0.07550.0175
2050.09350.06730.0902
2060.0369-0.00160.0247
2070.07700.01280.1342
208-0.0433-0.05080.0714
209-0.0781-0.0630-0.0208
2100.07490.0870-0.0056
2110.00000.02580.0703
2120.00210.0112-0.0394
2130.0277-0.0566-0.0602
2140.0096-0.01510.0742
215-0.00280.0153-0.0386
216-0.0336-0.1067-0.0550
217-0.0092-0.03900.0312
2180.04830.05870.1064
219-0.0696-0.0526-0.0316
220-0.0374-0.0245-0.0508
2210.02380.01590.0037
2220.08280.17900.0370
2230.12210.11950.1217
224-0.1046-0.1167-0.0420
225-0.0380-0.0383-0.0383
2260.07860.07870.0517
2270.00200.05550.1240
228-0.0358-0.0191-0.0686
229-0.0297-0.0231-0.0504
230-0.0765-0.07700.0790
231-0.0728-0.07520.0759
2320.02440.01530.0778
2330.06350.0659-0.0939
234-0.0605-0.0419-0.0390
2350.19860.17750.1089
236-0.0672-0.0515-0.0395
237-0.02750.0118-0.0469
238-0.0786-0.0884-0.0504
2390.01100.0121-0.0475
2400.03560.01810.0487
2410.03900.01960.0596
242-0.0685-0.07730.0046
243-0.0866-0.0880-0.0591
244-0.0034-0.0360-0.0489
2450.0223-0.02110.0323
2460.10250.1711-0.0372
2470.02510.06800.1266
248-0.0759-0.0612-0.0526
249-0.0157-0.0160-0.0638
Source: P3 source-disjoint test split · p3_02000795 · SHA-256 7102cb4fbccb5806…
p02Speech + burstsP3 curated burst test

Anger + 2 known burst(s)

Synthetic speech with inserted bursts · P3 synthetic speech/burst mixture · 12.0 s

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Triumph
z +3.74 / +4.63raw 1.94 / 2.40
Malevolence Malice
z +3.87 / +4.56raw 2.12 / 2.49
Pride
z +3.02 / +3.74raw 1.54 / 1.90
Contempt
z +2.56 / +3.14raw 1.19 / 1.46
Anger
z +2.19 / +2.96raw 1.34 / 1.82
Jealousy and Envy
z +1.47 / +2.60raw 0.71 / 1.21
Bitterness
z +2.53 / +2.50raw 1.07 / 1.05

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Tension (TENS)
z +2.23 / +2.78raw 3.45 / 3.92
Roughness (ROUG)
z +1.90 / +2.47raw 3.72 / 4.22
Dramatic style (S_DRAM)
z +1.93 / +2.23raw 4.94 / 5.28
Focus (FOCS)
z +1.86 / +2.11raw 4.39 / 4.67
Metallic character (METL)
z +1.66 / +2.08raw 3.21 / 3.61
Content appropriateness (EXPL)
z +1.66 / +2.05raw 0.62 / 0.75
Cartoonish style (S_CART)
z +1.57 / +1.77raw 3.77 / 4.04

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z +0.10 / +0.04raw 2.49 / 2.40
Vocal-burst blend (0–10)z -0.35 / -0.39raw 5.73 / 5.61
Content enjoyment (CE)z -0.41 / -0.49raw 4.38 / 4.30
Production quality (PQ)z -1.40 / -1.62raw 5.81 / 5.62
Speech signal quality, calibratedz -3.94 / -4.30raw 1.52 / 1.34
Background quality, calibratedz -2.27 / -2.49raw 1.28 / 1.08
Overall quality, calibratedz -2.70 / -2.86raw 1.20 / 1.11
P.808 MOS estimatez -1.03 / -1.16raw 2.91 / 2.84
score content enjoymentz -0.08 / +0.21raw 4.48 / 4.57
score overall qualityz -2.39 / -2.31raw 1.44 / 1.48
Background Noisez +0.35 / +0.45raw 0.90 / 0.93
Tempo (TEMP)z +0.71 / +1.05raw 2.96 / 3.37
Arousal (AROU)z +0.68 / +0.91raw 3.15 / 3.40
Valence (VALN)z +0.53 / +0.07raw 2.92 / 2.41
Recording quality (RCQL)z -1.54 / -1.64raw 1.39 / 1.30
Background noise (BKGN)z -0.87 / -0.85raw 2.04 / 2.06
talking speedz -0.30 / +0.18raw 7.88 / 9.71
Characters per second (CPS): 9.27 raw · z +0.06

P3 metadata teacher estimate; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8000.826
Identity · 250D0.7320.750

Vocal-burst regions

Known
Tiny
Base
0 s12.0 s

Purple marks known inserted burst intervals; other vocal sounds may be unlabeled.

Tiny: 3 regions, count-head estimate 1.78; Base: 2 regions, count-head estimate 1.43.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
Tiny15.405.760.36
Tiny26.026.360.34
Tiny311.1611.220.06
Base15.445.760.32
Base26.026.340.32

Known inserted intervals: 5.44–5.72 s, 6.03–6.35 s

All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.218+0.37-0.035-0.070.0070.570
1Amusementemo_Amusement0.368+0.420.402+0.45-0.0010.885
2Angeremo_Anger1.344+2.191.820+2.960.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.320+0.550.222+0.380.0050.572
4Aweemo_Awe0.164+0.380.112+0.260.0000.427
5Bitternessemo_Bitterness1.065+2.531.052+2.500.0130.415
6Concentrationemo_Concentration1.333+0.941.264+0.850.6340.741
7Confusionemo_Confusion0.073+0.13-0.051-0.100.0010.540
8Contemplationemo_Contemplation0.466+0.640.168+0.140.0820.598
9Contemptemo_Contempt1.189+2.561.456+3.14-0.0000.464
10Contentmentemo_Contentment0.266+0.52-0.037-0.070.0000.514
11Disappointmentemo_Disappointment0.593+0.850.309+0.440.0000.700
12Disgustemo_Disgust1.073+2.181.046+2.120.0070.489
13Distressemo_Distress0.560+0.680.501+0.610.0000.821
14Doubtemo_Doubt0.137+0.280.025+0.040.0060.466
15Elationemo_Elation0.646+0.870.859+1.160.0120.728
16Embarrassmentemo_Embarrassment-0.015-0.14-0.025-0.160.0430.430
17Emotional Numbnessemo_Emotional_Numbness0.165+0.180.437+0.470.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion-0.034-0.12-0.013-0.080.0280.539
19Fearemo_Fear0.287+0.450.322+0.500.0150.610
20Helplessnessemo_Helplessness0.103+0.15-0.009-0.01-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.870+0.991.046+1.220.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.785+0.901.209+1.500.1490.705
23Infatuationemo_Infatuation0.396+0.770.261+0.52-0.0150.533
24Interestemo_Interest2.535+1.042.435+0.931.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness0.234+0.460.301+0.58-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.709+1.471.211+2.600.0590.442
27Longingemo_Longing0.823+1.160.378+0.460.0870.632
28Malevolence Maliceemo_Malevolence_Malice2.121+3.872.489+4.560.0550.534
29Painemo_Pain0.054+0.100.020+0.04-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.590+0.930.710+1.130.0170.614
31Prideemo_Pride1.536+3.021.902+3.740.0020.509
32Reliefemo_Relief0.040+0.010.005-0.040.0330.681
33Sadnessemo_Sadness0.382+0.500.174+0.230.0000.759
34Sexual Lustemo_Sexual_Lust0.799+1.561.161+2.290.0300.494
35Shameemo_Shame0.050+0.070.016+0.010.0130.510
36Sournessemo_Sourness0.799+1.830.891+2.04-0.0050.439
37Teasingemo_Teasing0.415+0.760.507+0.930.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.214+0.250.061+0.030.0410.697
39Triumphemo_Triumph1.942+3.742.404+4.63-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg2.754-0.192.748-0.192.9841.238
41Arousal (AROU)vn_AROU_reg3.145+0.683.405+0.912.3591.147
42Arousal shift (ARSH)vn_ARSH_reg3.505+1.073.214+0.732.5800.867
43Vocal attack (ATCK)vn_ATCK_reg2.715+0.433.322+0.902.1621.295
44Background noise (BKGN)vn_BKGN_reg2.040-0.872.058-0.852.7030.761
45Brightness (BRGT)vn_BRGT_reg0.824-1.440.925-1.311.9910.813
46Speech chunking (CHNK)vn_CHNK_reg2.225+0.312.211+0.291.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg0.873-1.210.828-1.262.0180.945
48Cognitive load (COGL)vn_COGL_reg1.210-0.831.177-0.862.0941.061
49Dynamic arc (DARC)vn_DARC_reg3.039+1.033.581+1.421.6151.380
50Disfluency (DFLU)vn_DFLU_reg1.844-0.542.104-0.322.4881.184
51Emphasis (EMPH)vn_EMPH_reg3.424+0.544.056+0.952.6071.524
52Esthetics (ESTH)vn_ESTH_reg1.811-0.761.701-0.902.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.623+1.660.746+2.050.0860.322
54Focus (FOCS)vn_FOCS_reg4.392+1.864.669+2.112.3441.102
55Fullness (FULL)vn_FULL_reg0.128-1.860.329-1.631.7780.889
56Perceived gender (GEND)vn_GEND_reg4.584+0.335.051+0.634.0551.588
57Harmonicity (HARM)vn_HARM_reg1.437-1.251.262-1.462.4800.836
58Metallic character (METL)vn_METL_reg3.210+1.663.614+2.081.6210.958
59Pitch range (RANG)vn_RANG_reg3.780+0.963.943+1.152.9140.897
60Recording quality (RCQL)vn_RCQL_reg1.387-1.541.296-1.642.7540.888
61Vocal register (REGS)vn_REGS_reg1.064-0.110.824-0.301.1981.268
62Respiration (RESP)vn_RESP_reg3.511+0.793.858+1.122.6801.056
63Roughness (ROUG)vn_ROUG_reg3.717+1.904.220+2.472.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg2.224+0.112.120-0.022.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg1.761-0.471.526-0.762.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg1.359-0.551.397-0.511.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg1.069-0.440.860-0.681.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg2.753-0.042.747-0.042.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg1.536-0.871.684-0.672.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg2.071+0.002.200+0.162.0700.803
71Smoothness (SMTH)vn_SMTH_reg1.693-0.411.552-0.542.1431.093
72Stance (STNC)vn_STNC_reg3.988+1.194.024+1.222.6331.137
73Speech structure (STRU)vn_STRU_reg1.674-0.481.463-0.702.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg1.207-0.120.978-0.281.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg2.591+1.442.526+1.381.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg3.765+1.574.043+1.771.5601.401
77Casual style (S_CASU)vn_S_CASU_reg4.326+0.424.508+0.563.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg-0.363-2.24-0.057-2.012.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg4.942+1.935.280+2.232.7211.148
80Formal style (S_FORM)vn_S_FORM_reg0.261-0.920.001-1.171.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg1.498-1.031.304-1.182.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg0.270-0.720.471-0.581.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.238-0.300.299-0.190.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg2.906+0.422.888+0.412.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg2.591+1.073.039+1.421.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg4.247+1.294.302+1.332.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg-0.908-1.27-1.194-1.500.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg2.672+0.672.254+0.411.5971.618
89Tempo (TEMP)vn_TEMP_reg2.964+0.713.372+1.052.1171.190
90Tension (TENS)vn_TENS_reg3.454+2.233.922+2.781.5590.852
91Valence (VALN)vn_VALN_reg2.919+0.532.407+0.072.3281.118
92Valence shift (VALS)vn_VALS_reg2.602+0.082.549+0.032.5181.079
93Velocity flux (VFLX)vn_VFLX_reg2.546+0.243.032+0.642.2561.208
94Volatility (VOLT)vn_VOLT_reg3.885+1.474.089+1.702.5860.886
95Vulnerability (VULN)vn_VULN_reg0.651-2.030.434-2.252.6480.984
96Warmth (WARM)vn_WARM_reg1.439-1.251.305-1.402.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_62.494+0.102.395+0.042.3321.621
98Vocal-burst blend (0–10)blend_0_105.726-0.355.614-0.396.6762.738
99Legacy R_quality (no P3 test labels)R_quality1.009-0.250.989-0.311.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.254+0.013.122-0.443.2500.293
101Arousaleiv_extra:Arousal2.703+3.252.901+3.701.2890.435
102Authenticityeiv_extra:Authenticity2.718+0.332.604-0.412.6670.153
103Background Noiseeiv_extra:Background_Noise0.901+0.350.928+0.450.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.095+0.181.015-0.081.0400.303
105Gendereiv_extra:Gender-0.391+0.23-0.732-0.11-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched2.042+0.862.024+0.791.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive3.039+2.513.017+2.461.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.231+0.262.225+0.242.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous0.970+0.120.974+0.130.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.109-0.54-0.108-1.220.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant1.267+2.751.490+3.440.3760.324
112Valenceeiv_extra:Valence-0.101-0.28-0.273-0.730.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.459-0.581.526-0.431.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold-0.308-1.20-0.369-1.400.0510.300
115score background qualityeiv_extra:score_background_quality1.795-2.171.806-2.163.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.483-0.084.570+0.214.5070.303
117score overall qualityeiv_extra:score_overall_quality1.442-2.391.482-2.312.6520.507
118score speech qualityeiv_extra:score_speech_quality1.569-1.641.562-1.691.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE4.380-0.414.297-0.494.7991.018
120Content usefulness (CU)audiobox:CU4.177-1.474.286-1.365.6491.001
121Production complexity (PC)audiobox:PC4.786+1.524.111+1.102.3311.617
122Production quality (PQ)audiobox:PQ5.810-1.405.618-1.627.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw1.361-3.791.152-4.113.9000.669
124Background quality, rawdnsmos:BAK_raw1.160-1.820.984-1.983.2111.124
125Overall quality, rawdnsmos:OVRL_raw1.096-2.451.007-2.563.0850.812
126Speech signal quality, calibrateddnsmos:SIG1.516-3.941.336-4.303.4880.501
127Background quality, calibrateddnsmos:BAK1.282-2.271.079-2.493.3890.928
128Overall quality, calibrateddnsmos:OVRL1.202-2.701.108-2.862.8400.606
129P.808 MOS estimatednsmos:P808_MOS2.908-1.032.844-1.163.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p1.021+1.580.889+1.370.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.891-0.100.997+0.090.9470.554
132Agevoiceclap_attribute:Age2.910-0.582.802-0.783.2290.547
133Amusementvoiceclap_attribute:Amusement0.969+1.011.050+1.150.3920.572
134Angervoiceclap_attribute:Anger1.798+1.592.095+2.180.9920.506
135Arousalvoiceclap_attribute:Arousal2.992+2.532.992+2.531.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.833+0.681.096+1.230.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.694-0.462.701-0.412.7700.167
138Awevoiceclap_attribute:Awe0.740+0.770.795+0.990.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.868+1.140.932+1.380.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.782+0.800.883+1.230.5920.236
141Concentrationvoiceclap_attribute:Concentration0.944-0.510.787-0.801.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.098+0.061.196+0.271.0690.468
143Confusionvoiceclap_attribute:Confusion0.702+0.180.735+0.270.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.450-0.810.514-0.660.7800.406
145Contemptvoiceclap_attribute:Contempt1.290+1.721.269+1.650.7090.339
146Contentmentvoiceclap_attribute:Contentment0.512-0.250.458-0.420.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.944+0.631.024+0.910.7600.291
148Disgustvoiceclap_attribute:Disgust0.908+1.371.043+2.020.6250.207
149Distressvoiceclap_attribute:Distress1.077+0.431.192+0.760.9270.350
150Doubtvoiceclap_attribute:Doubt0.359-0.930.334-1.070.5350.188
151Elationvoiceclap_attribute:Elation1.576+2.611.639+2.770.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.365+0.180.344+0.120.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.424-0.900.489-0.660.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.262-0.770.405-0.220.4620.261
155Fearvoiceclap_attribute:Fear0.670+0.240.738+0.570.6180.212
156Gendervoiceclap_attribute:Gender-0.294-0.14-0.433-0.28-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.374-1.060.444-0.800.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched2.111+0.722.085+0.651.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism2.208+3.802.127+3.590.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability2.200+2.122.577+2.821.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.877+0.810.843+0.710.6080.332
162Interestvoiceclap_attribute:Interest2.117+1.182.012+0.941.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness1.083+1.251.340+1.810.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.859+1.380.848+1.330.5230.245
165Longingvoiceclap_attribute:Longing0.526-0.050.562+0.110.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice1.663+2.401.663+2.400.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive3.224+2.443.261+2.501.7750.594
168Painvoiceclap_attribute:Pain0.395+0.060.455+0.340.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy1.029+1.751.022+1.720.5030.301
170Pridevoiceclap_attribute:Pride1.314+1.531.412+1.810.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.279-0.102.142-0.522.3130.328
172Reliefvoiceclap_attribute:Relief0.507-1.250.527-1.201.0250.415
173Sadnessvoiceclap_attribute:Sadness0.462-0.930.495-0.820.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous1.059+0.451.320+0.910.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust1.211+1.371.247+1.460.6350.419
176Shamevoiceclap_attribute:Shame0.412-0.400.398-0.460.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh-0.507-1.85-0.611-2.140.1600.360
178Sournessvoiceclap_attribute:Sourness0.951+0.850.960+0.880.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant1.579+2.671.736+3.100.6230.359
180Teasingvoiceclap_attribute:Teasing1.029+1.291.079+1.420.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.917+1.011.799+0.781.3840.529
182Triumphvoiceclap_attribute:Triumph1.662+2.631.785+2.980.7160.359
183Valencevoiceclap_attribute:Valence0.010+0.03-0.256-0.39-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached1.318-0.581.403-0.421.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold-0.260-0.48-0.370-0.76-0.0660.400
186durationvoiceclap_attribute:duration11.657+0.1211.147+0.0410.9146.354
187score background qualityvoiceclap_attribute:score_background_quality2.475-1.542.510-1.473.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.716+0.414.742+0.584.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.037-1.452.056-1.402.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.827-0.931.840-0.731.8910.069
191talking speedvoiceclap_attribute:talking_speed7.884-0.309.714+0.189.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.00130.0433-0.0739
1-0.0136-0.0879-0.0739
20.05260.04490.1046
3-0.0581-0.0490-0.0124
40.02560.0371-0.0196
5-0.1324-0.1011-0.0941
60.08420.1156-0.0039
7-0.0820-0.03830.0127
8-0.0142-0.0172-0.0557
9-0.01470.0015-0.0498
100.23090.28770.2683
110.01850.07570.0417
120.00450.02040.0180
13-0.0621-0.0399-0.0181
140.18820.12000.1236
15-0.1353-0.1193-0.0666
16-0.1574-0.1382-0.0920
17-0.0868-0.0911-0.0773
18-0.02760.03450.0870
19-0.0838-0.1074-0.0703
20-0.0879-0.0678-0.0744
21-0.0739-0.0738-0.0684
220.02340.02610.0812
230.02460.06370.0675
24-0.0375-0.0587-0.0311
25-0.0893-0.1005-0.0652
26-0.02980.00280.0211
27-0.1176-0.0911-0.0693
28-0.00220.03360.0931
29-0.0194-0.0559-0.0059
30-0.00120.0294-0.0800
31-0.0860-0.0437-0.0686
320.17430.19720.0774
33-0.0599-0.04370.0387
340.0248-0.00510.0834
35-0.03380.0064-0.0659
36-0.0174-0.0540-0.0423
37-0.1154-0.0619-0.0444
380.07760.05020.0151
39-0.0805-0.0458-0.0637
40-0.1500-0.0873-0.0717
41-0.01310.00370.0079
42-0.0301-0.0272-0.0750
43-0.0978-0.0961-0.0716
44-0.1088-0.0678-0.0714
45-0.0041-0.02170.0051
46-0.00120.0192-0.0701
47-0.0563-0.0729-0.0115
48-0.1088-0.0779-0.0587
49-0.0636-0.0769-0.0751
50-0.0452-0.0253-0.0760
510.16540.16200.1561
52-0.0296-0.05150.0080
53-0.0923-0.0930-0.0757
54-0.0581-0.0659-0.0356
55-0.0489-0.0462-0.0130
560.01190.0543-0.0488
57-0.0555-0.0854-0.0283
58-0.0732-0.0449-0.0430
59-0.1037-0.0779-0.0745
60-0.0439-0.03640.0175
61-0.0151-0.0388-0.0684
620.0281-0.0101-0.0028
63-0.0923-0.0663-0.0706
640.09010.10800.0431
65-0.0309-0.0421-0.0550
660.0212-0.0675-0.1028
670.15960.13190.0508
680.02440.0389-0.0019
690.17940.16460.1873
70-0.0160-0.03050.0731
710.10510.16960.1891
720.25290.23870.2201
73-0.0128-0.01080.0094
74-0.0511-0.0625-0.0804
750.06690.02940.0139
760.11180.09040.1998
770.03110.06310.1659
780.05810.08980.0537
790.05880.07570.1925
800.10370.12750.0715
810.20000.21610.1992
820.05110.08350.1165
830.00560.06690.0662
84-0.1228-0.0904-0.0921
850.01250.01460.0434
860.02570.00880.1166
87-0.0008-0.03190.0468
88-0.0614-0.0421-0.0690
89-0.03820.0067-0.0120
900.10370.09170.1327
910.02650.04180.0115
920.0161-0.0110-0.0211
93-0.1000-0.1168-0.0726
940.04030.01770.0035
950.05110.0115-0.0099
96-0.0313-0.02970.0579
97-0.00300.02870.0735
98-0.0540-0.0751-0.0751
990.23680.18720.1240
1000.06540.04430.0477
1010.20290.24620.1376
102-0.0956-0.0754-0.0795
1030.05510.02580.0178
104-0.0311-0.0128-0.0727
105-0.0252-0.0062-0.0199
106-0.02250.00080.1113
1070.04960.06440.0766
1080.04470.01980.0457
109-0.0162-0.00690.0014
1100.07980.08980.0048
1110.10590.10870.2241
112-0.0178-0.01700.1010
113-0.0684-0.0754-0.1046
1140.27940.32030.2737
1150.15510.12750.1503
116-0.0563-0.0688-0.0637
117-0.02920.00380.0119
118-0.0691-0.0581-0.0723
119-0.1007-0.0816-0.0718
120-0.0309-0.0366-0.0764
121-0.0300-0.0757-0.0476
122-0.0592-0.0394-0.0764
123-0.0629-0.0606-0.0394
1240.04890.0023-0.0269
125-0.1015-0.0936-0.0741
1260.0052-0.00370.0838
127-0.0351-0.0729-0.0208
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
00.09130.10210.0930
1-0.0703-0.0704-0.0417
20.01210.0464-0.0412
3-0.0136-0.0513-0.0284
4-0.0389-0.0196-0.0392
5-0.01780.00870.0295
60.09820.13390.0870
7-0.0298-0.0149-0.0100
8-0.02470.0053-0.0765
90.04450.09270.1493
10-0.0342-0.0486-0.0453
110.02880.0436-0.0018
12-0.0578-0.04410.0245
130.15460.15760.1284
14-0.0907-0.0917-0.0616
150.10000.09420.1400
16-0.1250-0.1031-0.0698
17-0.0058-0.0332-0.0518
180.04800.05300.1569
19-0.0453-0.0283-0.0684
200.0371-0.0254-0.0432
21-0.0488-0.0439-0.0400
22-0.0101-0.0431-0.0180
23-0.0907-0.0709-0.0417
24-0.0123-0.0362-0.0368
25-0.0607-0.0759-0.0793
260.12560.14870.0190
27-0.0327-0.0231-0.0351
280.0301-0.00030.0504
29-0.0584-0.0459-0.0415
30-0.0102-0.0347-0.0375
310.03810.02280.1704
320.08140.09470.0432
330.08490.14080.0859
34-0.1041-0.0957-0.0515
350.10930.0932-0.0001
36-0.0407-0.04190.0437
37-0.0082-0.0244-0.0298
38-0.0181-0.0059-0.0166
39-0.1041-0.1190-0.0643
40-0.0692-0.0617-0.0500
41-0.0631-0.0419-0.0495
42-0.0410-0.03050.0147
430.04040.0089-0.0485
440.0209-0.0053-0.0519
450.03120.01090.0035
46-0.0217-0.0525-0.0634
470.10000.09470.0350
48-0.0058-0.0496-0.0461
49-0.0421-0.0402-0.0050
500.11280.09670.0976
51-0.0311-0.0178-0.0340
52-0.0424-0.0389-0.0425
53-0.0509-0.0306-0.0417
54-0.0439-0.0446-0.0231
550.07730.07830.0323
56-0.0512-0.0615-0.0593
57-0.00950.00850.0220
58-0.0266-0.0112-0.0117
590.11450.08970.0952
60-0.0692-0.0813-0.0452
610.13310.1497-0.0401
62-0.0936-0.0644-0.0379
63-0.01130.0088-0.0469
64-0.0616-0.0607-0.0480
65-0.0355-0.0322-0.0526
660.10230.08770.1108
670.0404-0.0023-0.0032
680.0033-0.01340.0090
69-0.0837-0.1096-0.0797
70-0.0375-0.0491-0.0343
71-0.0253-0.0315-0.0452
72-0.0677-0.04210.0764
73-0.0872-0.1006-0.0770
74-0.0831-0.0674-0.0465
750.13720.10810.1122
760.09770.09520.1376
77-0.02280.01750.0354
780.02440.01410.0371
79-0.0427-0.0224-0.0461
80-0.0663-0.0421-0.0458
81-0.0477-0.0605-0.0516
820.12560.13090.1698
83-0.0622-0.0568-0.0383
840.02080.04070.0513
85-0.0187-0.0488-0.0461
86-0.0273-0.0411-0.0411
87-0.0075-0.0191-0.0367
880.02080.00570.0286
89-0.0202-0.0580-0.0417
900.02560.0394-0.0457
91-0.0427-0.0335-0.0412
92-0.0049-0.01070.0531
93-0.0613-0.0627-0.0523
94-0.0453-0.0858-0.0746
950.06050.13780.0181
96-0.0738-0.0573-0.0412
970.05000.03500.0999
980.07620.04560.0580
99-0.0773-0.0654-0.0576
1000.05380.06220.1366
101-0.0459-0.0407-0.0435
102-0.0625-0.0345-0.0366
103-0.1017-0.0987-0.0761
104-0.0488-0.0102-0.0600
1050.04070.0426-0.0599
106-0.0671-0.0664-0.0355
107-0.0353-0.0217-0.0413
108-0.0488-0.0513-0.0343
1090.0017-0.0058-0.0443
110-0.0416-0.0066-0.0205
1110.12320.09070.0460
1120.01040.04020.0428
113-0.0315-0.0739-0.0786
114-0.0767-0.0654-0.0642
115-0.0240-0.0649-0.0326
1160.06280.09070.0450
117-0.1232-0.1190-0.0751
118-0.0860-0.0714-0.0601
1190.04270.05230.0248
1200.01630.0228-0.0247
1210.03780.04360.0730
122-0.0924-0.0863-0.0435
1230.0279-0.0094-0.0476
124-0.0085-0.0274-0.0214
125-0.0558-0.0449-0.0448
1260.10460.13480.1593
127-0.0324-0.0330-0.0474
128-0.0602-0.0763-0.0574
129-0.0219-0.04390.0078
130-0.0549-0.0061-0.0323
131-0.0416-0.0399-0.0403
132-0.0238-0.0219-0.0189
133-0.0401-0.0456-0.0455
1340.08890.08180.0174
1350.00450.01480.0587
136-0.0211-0.0568-0.0353
137-0.0791-0.0632-0.0428
138-0.0517-0.0555-0.0406
139-0.0503-0.0783-0.0471
140-0.00310.0399-0.0288
141-0.0639-0.0449-0.0456
142-0.0555-0.0530-0.0398
143-0.0195-0.0030-0.0862
1440.00130.0429-0.0480
145-0.00970.00590.0047
1460.13080.11200.1117
1470.07500.10010.0393
1480.08950.03940.0912
149-0.0622-0.0622-0.0593
150-0.0497-0.0468-0.0453
1510.06050.0555-0.0525
1520.0124-0.00080.0778
1530.14420.09170.0193
1540.07150.05870.0124
155-0.0055-0.0617-0.0601
156-0.0246-0.0754-0.0461
1570.08200.05970.0365
158-0.0314-0.03200.0384
159-0.1104-0.0763-0.0539
1600.05580.05600.1549
161-0.1110-0.1497-0.1008
1620.07090.11850.0405
163-0.0503-0.0617-0.0513
164-0.0436-0.0538-0.0404
165-0.0913-0.0739-0.0232
166-0.0199-0.0092-0.0536
167-0.1093-0.1165-0.0584
1680.04820.01150.0266
1690.00340.03370.0211
1700.02730.03270.1466
171-0.0121-0.0138-0.0405
1720.07440.07880.0352
1730.02910.0213-0.0354
174-0.0182-0.0233-0.0371
175-0.0497-0.0873-0.0667
176-0.0674-0.0719-0.0392
177-0.0041-0.00190.0563
178-0.0514-0.0513-0.0409
1790.03630.0219-0.0113
180-0.0392-0.0449-0.0527
181-0.00400.0191-0.0179
182-0.0401-0.0516-0.0398
183-0.0430-0.0325-0.0469
184-0.0301-0.0320-0.0468
185-0.0994-0.0749-0.0551
1860.0407-0.01850.0472
187-0.0526-0.0813-0.0524
1880.04650.06400.0405
189-0.0283-0.0327-0.0730
1900.10520.10510.0811
191-0.0102-0.00060.0265
1920.03870.02320.0409
1930.05640.04780.1102
194-0.0526-0.04390.0435
195-0.0590-0.0602-0.0437
1960.03460.00140.0188
197-0.0520-0.0436-0.0369
198-0.0523-0.0523-0.0445
199-0.0471-0.0387-0.0506
200-0.0167-0.0186-0.0090
201-0.0497-0.0273-0.0434
202-0.1006-0.1398-0.0958
203-0.0163-0.0350-0.0574
2040.14420.14080.1459
2050.06100.04160.1215
206-0.0375-0.02370.0526
2070.00760.04040.0043
208-0.0671-0.0724-0.0478
209-0.00660.00900.0319
2100.05640.03000.0637
211-0.0276-0.0073-0.0354
2120.03600.0364-0.0008
2130.16390.12050.0663
2140.04880.02580.0256
215-0.0014-0.0006-0.0387
2160.11630.10810.0686
2170.00500.04340.0211
2180.14360.12390.0928
219-0.0837-0.0295-0.0553
220-0.0288-0.0024-0.0137
2210.02910.0175-0.0399
222-0.0584-0.0374-0.0540
223-0.0410-0.0550-0.0369
2240.08660.07930.2059
2250.0026-0.0285-0.0383
2260.07500.04680.0052
227-0.0605-0.0714-0.0608
2280.05000.0190-0.0005
2290.01140.00970.0132
230-0.00280.0115-0.0469
2310.03490.07490.0408
232-0.0185-0.0374-0.0299
2330.07410.06220.0567
234-0.0421-0.0352-0.0391
235-0.0082-0.00650.0339
236-0.0465-0.0543-0.0395
2370.01860.0086-0.0092
238-0.0439-0.0496-0.0505
2390.0005-0.0058-0.0259
240-0.0225-0.0399-0.0469
2410.04420.04290.0869
242-0.0953-0.1041-0.0678
2430.06630.11900.2343
244-0.0027-0.0085-0.0490
245-0.0228-0.03320.0566
246-0.0674-0.0397-0.0373
2470.19420.13680.1653
248-0.0025-0.02730.0230
2490.06800.03200.0079
Source: P3 source-disjoint test split · p3_02007119 · SHA-256 ed8104e92c44a469…
p03Speech + burstsP3 curated burst test

Astonishment Surprise + 3 known burst(s)

Synthetic speech with inserted bursts · P3 synthetic speech/burst mixture · 8.0 s

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Astonishment Surprise
z +2.85 / +3.16raw 1.63 / 1.81
Disgust
z +2.22 / +1.76raw 1.10 / 0.87
Teasing
z +1.95 / +1.86raw 1.06 / 1.01
Relief
z +0.70 / +1.47raw 0.51 / 1.03
Anger
z +0.96 / +1.41raw 0.59 / 0.87
Malevolence Malice
z +1.08 / +1.14raw 0.63 / 0.66
Contempt
z +1.07 / +0.87raw 0.49 / 0.40

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Chest resonance (R_CHST)
z +2.24 / +2.27raw 4.02 / 4.05
Warmth (WARM)
z +2.18 / +1.99raw 4.56 / 4.39
Narrator style (S_NARR)
z +1.57 / +2.15raw 3.57 / 4.41
Roughness (ROUG)
z +1.39 / +1.85raw 3.26 / 3.68
Content appropriateness (EXPL)
z +1.84 / +1.23raw 0.68 / 0.48
Volatility (VOLT)
z +0.97 / +1.73raw 3.44 / 4.12
Storytelling style (S_STRY)
z +1.25 / +1.61raw 4.19 / 4.68

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z -0.48 / -0.32raw 1.56 / 1.81
Vocal-burst blend (0–10)z -2.11 / -2.55raw 0.89 / -0.30
Content enjoyment (CE)z +0.58 / +0.54raw 5.39 / 5.35
Production quality (PQ)z +1.14 / +1.01raw 8.04 / 7.93
Speech signal quality, calibratedz +0.11 / -0.06raw 3.54 / 3.46
Background quality, calibratedz +0.56 / +0.40raw 3.91 / 3.76
Overall quality, calibratedz +0.59 / +0.35raw 3.20 / 3.05
P.808 MOS estimatez +0.22 / +0.29raw 3.53 / 3.56
score content enjoymentz +0.47 / +0.45raw 4.65 / 4.64
score overall qualityz +0.62 / +0.65raw 2.97 / 2.98
Background Noisez -0.49 / -0.32raw 0.68 / 0.72
Tempo (TEMP)z -0.19 / -0.19raw 1.89 / 1.89
Arousal (AROU)z -0.39 / -0.02raw 1.91 / 2.33
Valence (VALN)z -0.34 / -0.35raw 1.95 / 1.93
Recording quality (RCQL)z +0.34 / +0.69raw 3.06 / 3.37
Background noise (BKGN)z +0.50 / +1.29raw 3.09 / 3.69
talking speedz +0.30 / +0.37raw 10.14 / 10.42
Characters per second (CPS): 9.16 raw · z +0.03

P3 metadata teacher estimate; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8750.887
Identity · 250D0.8120.819

Vocal-burst regions

Known
Tiny
Base
0 s8.0 s

Purple marks known inserted burst intervals; other vocal sounds may be unlabeled.

Tiny: 3 regions, count-head estimate 4.02; Base: 3 regions, count-head estimate 4.36.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
Tiny10.020.500.48
Tiny25.766.340.58
Tiny36.787.720.94
Base10.020.500.48
Base25.746.340.60
Base36.787.700.92

Known inserted intervals: 5.76–6.33 s, 6.79–7.70 s, 0.02–0.47 s

All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.163+0.270.156+0.260.0070.570
1Amusementemo_Amusement0.939+1.060.922+1.04-0.0010.885
2Angeremo_Anger0.591+0.960.871+1.410.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise1.632+2.851.814+3.160.0050.572
4Aweemo_Awe0.090+0.210.070+0.160.0000.427
5Bitternessemo_Bitterness0.069+0.140.011-0.000.0130.415
6Concentrationemo_Concentration0.237-0.540.217-0.560.6340.741
7Confusionemo_Confusion0.218+0.400.272+0.500.0010.540
8Contemplationemo_Contemplation0.020-0.100.027-0.090.0820.598
9Contemptemo_Contempt0.494+1.070.404+0.87-0.0000.464
10Contentmentemo_Contentment-0.012-0.020.012+0.020.0000.514
11Disappointmentemo_Disappointment0.142+0.200.090+0.130.0000.700
12Disgustemo_Disgust1.096+2.220.870+1.760.0070.489
13Distressemo_Distress0.173+0.210.216+0.260.0000.821
14Doubtemo_Doubt0.257+0.540.181+0.380.0060.466
15Elationemo_Elation0.484+0.650.689+0.930.0120.728
16Embarrassmentemo_Embarrassment0.186+0.330.241+0.460.0430.430
17Emotional Numbnessemo_Emotional_Numbness-0.013-0.02-0.056-0.070.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.035+0.010.050+0.040.0280.539
19Fearemo_Fear0.376+0.590.350+0.550.0150.610
20Helplessnessemo_Helplessness0.002+0.010.011+0.02-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.314+0.260.336+0.290.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.631+0.680.765+0.870.1490.705
23Infatuationemo_Infatuation0.260+0.520.105+0.22-0.0150.533
24Interestemo_Interest1.697+0.121.606+0.021.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness0.032+0.100.019+0.08-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.130+0.160.106+0.110.0590.442
27Longingemo_Longing0.126+0.06-0.042-0.200.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.634+1.080.665+1.140.0550.534
29Painemo_Pain0.136+0.240.104+0.18-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.484+0.760.564+0.890.0170.614
31Prideemo_Pride0.066+0.130.108+0.210.0020.509
32Reliefemo_Relief0.513+0.701.032+1.470.0330.681
33Sadnessemo_Sadness-0.040-0.050.035+0.050.0000.759
34Sexual Lustemo_Sexual_Lust0.328+0.600.221+0.390.0300.494
35Shameemo_Shame0.073+0.120.059+0.090.0130.510
36Sournessemo_Sourness0.368+0.850.218+0.51-0.0050.439
37Teasingemo_Teasing1.064+1.951.014+1.860.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.166+0.180.365+0.470.0410.697
39Triumphemo_Triumph0.249+0.490.509+0.99-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg3.048+0.053.186+0.162.9841.238
41Arousal (AROU)vn_AROU_reg1.911-0.392.331-0.022.3591.147
42Arousal shift (ARSH)vn_ARSH_reg2.800+0.253.710+1.302.5800.867
43Vocal attack (ATCK)vn_ATCK_reg1.396-0.591.427-0.572.1621.295
44Background noise (BKGN)vn_BKGN_reg3.086+0.503.686+1.292.7030.761
45Brightness (BRGT)vn_BRGT_reg2.007+0.022.225+0.291.9910.813
46Speech chunking (CHNK)vn_CHNK_reg2.391+0.442.159+0.251.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg1.754-0.281.941-0.082.0180.945
48Cognitive load (COGL)vn_COGL_reg1.350-0.701.153-0.892.0941.061
49Dynamic arc (DARC)vn_DARC_reg1.743+0.092.868+0.911.6151.380
50Disfluency (DFLU)vn_DFLU_reg2.063-0.362.261-0.192.4881.184
51Emphasis (EMPH)vn_EMPH_reg3.835+0.814.584+1.302.6071.524
52Esthetics (ESTH)vn_ESTH_reg2.736+0.432.799+0.512.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.679+1.840.482+1.230.0860.322
54Focus (FOCS)vn_FOCS_reg3.321+0.893.845+1.362.3441.102
55Fullness (FULL)vn_FULL_reg2.225+0.502.503+0.821.7780.889
56Perceived gender (GEND)vn_GEND_reg6.034+1.256.059+1.264.0551.588
57Harmonicity (HARM)vn_HARM_reg1.773-0.851.624-1.022.4800.836
58Metallic character (METL)vn_METL_reg1.233-0.401.549-0.071.6210.958
59Pitch range (RANG)vn_RANG_reg3.294+0.423.602+0.772.9140.897
60Recording quality (RCQL)vn_RCQL_reg3.057+0.343.370+0.692.7540.888
61Vocal register (REGS)vn_REGS_reg0.294-0.710.137-0.841.1981.268
62Respiration (RESP)vn_RESP_reg3.321+0.613.444+0.722.6801.056
63Roughness (ROUG)vn_ROUG_reg3.263+1.393.676+1.852.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg4.025+2.244.051+2.272.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg1.622-0.641.780-0.452.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg1.217-0.691.243-0.661.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg2.409+1.162.579+1.361.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg2.423-0.352.453-0.322.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg2.045-0.192.076-0.152.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg2.087+0.022.016-0.072.0700.803
71Smoothness (SMTH)vn_SMTH_reg2.822+0.622.744+0.552.1431.093
72Stance (STNC)vn_STNC_reg3.508+0.773.748+0.982.6331.137
73Speech structure (STRU)vn_STRU_reg1.706-0.451.665-0.492.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg0.911-0.330.943-0.301.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg2.108+0.982.445+1.301.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg2.589+0.733.124+1.121.5601.401
77Casual style (S_CASU)vn_S_CASU_reg3.292-0.393.073-0.563.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg3.920+1.014.104+1.152.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg3.763+0.913.972+1.092.7211.148
80Formal style (S_FORM)vn_S_FORM_reg0.726-0.460.718-0.471.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg2.907+0.063.034+0.162.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg3.572+1.574.413+2.151.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.293-0.200.239-0.300.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg3.132+0.593.650+0.992.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg2.571+1.052.346+0.881.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg4.193+1.254.683+1.612.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg0.669+0.020.988+0.280.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg2.413+0.502.541+0.581.5971.618
89Tempo (TEMP)vn_TEMP_reg1.891-0.191.892-0.192.1171.190
90Tension (TENS)vn_TENS_reg2.472+1.072.882+1.551.5590.852
91Valence (VALN)vn_VALN_reg1.948-0.341.935-0.352.3281.118
92Valence shift (VALS)vn_VALS_reg2.476-0.042.619+0.092.5181.079
93Velocity flux (VFLX)vn_VFLX_reg1.175-0.892.194-0.052.2561.208
94Volatility (VOLT)vn_VOLT_reg3.444+0.974.123+1.732.5860.886
95Vulnerability (VULN)vn_VULN_reg2.040-0.622.668+0.022.6480.984
96Warmth (WARM)vn_WARM_reg4.556+2.184.387+1.992.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_61.558-0.481.815-0.322.3321.621
98Vocal-burst blend (0–10)blend_0_100.890-2.11-0.300-2.556.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.386-2.100.298-2.361.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.255+0.013.224-0.093.2500.293
101Arousaleiv_extra:Arousal2.192+2.082.261+2.231.2890.435
102Authenticityeiv_extra:Authenticity2.847+1.182.909+1.592.6670.153
103Background Noiseeiv_extra:Background_Noise0.679-0.490.723-0.320.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.075+0.121.085+0.151.0400.303
105Gendereiv_extra:Gender-0.678-0.06-0.709-0.09-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.589-0.951.532-1.181.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive2.104+0.522.062+0.431.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.143-0.002.227+0.252.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous1.483+1.621.509+1.700.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.096-0.580.202-0.250.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.916+1.670.871+1.530.3760.324
112Valenceeiv_extra:Valence0.148+0.380.234+0.600.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.371-0.791.293-0.971.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold0.162+0.370.079+0.090.0510.300
115score background qualityeiv_extra:score_background_quality3.572+0.393.659+0.523.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.649+0.474.644+0.454.5070.303
117score overall qualityeiv_extra:score_overall_quality2.967+0.622.981+0.652.6520.507
118score speech qualityeiv_extra:score_speech_quality1.789-0.131.797-0.071.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.391+0.585.347+0.544.7991.018
120Content usefulness (CU)audiobox:CU6.643+0.996.698+1.055.6491.001
121Production complexity (PC)audiobox:PC2.404+0.052.571+0.152.3311.617
122Production quality (PQ)audiobox:PQ8.043+1.147.933+1.017.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw4.011+0.173.849-0.083.9000.669
124Background quality, rawdnsmos:BAK_raw3.982+0.693.815+0.543.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.627+0.673.427+0.423.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.545+0.113.456-0.063.4880.501
127Background quality, calibrateddnsmos:BAK3.908+0.563.760+0.403.3890.928
128Overall quality, calibrateddnsmos:OVRL3.200+0.593.053+0.352.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.530+0.223.563+0.293.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p1.614+2.491.680+2.600.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.686-0.470.708-0.430.9470.554
132Agevoiceclap_attribute:Age3.113-0.213.004-0.413.2290.547
133Amusementvoiceclap_attribute:Amusement0.955+0.981.097+1.230.3920.572
134Angervoiceclap_attribute:Anger1.620+1.241.766+1.530.9920.506
135Arousalvoiceclap_attribute:Arousal2.282+1.382.622+1.931.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise1.541+2.152.140+3.400.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.914+0.863.021+1.502.7700.167
138Awevoiceclap_attribute:Awe0.541+0.010.765+0.870.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.511-0.210.554-0.050.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.555-0.160.620+0.120.5920.236
141Concentrationvoiceclap_attribute:Concentration0.369-1.590.207-1.901.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant0.705-0.780.668-0.861.0690.468
143Confusionvoiceclap_attribute:Confusion0.721+0.230.795+0.450.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.271-1.250.069-1.750.7800.406
145Contemptvoiceclap_attribute:Contempt1.107+1.181.136+1.260.7090.339
146Contentmentvoiceclap_attribute:Contentment0.387-0.660.346-0.790.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.952+0.660.900+0.480.7600.291
148Disgustvoiceclap_attribute:Disgust1.078+2.191.185+2.710.6250.207
149Distressvoiceclap_attribute:Distress1.145+0.621.457+1.520.9270.350
150Doubtvoiceclap_attribute:Doubt0.529-0.030.520-0.080.5350.188
151Elationvoiceclap_attribute:Elation0.766+0.570.857+0.800.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.380+0.230.428+0.380.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.476-0.710.331-1.250.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.479+0.070.294-0.640.4620.261
155Fearvoiceclap_attribute:Fear0.905+1.351.036+1.970.6180.212
156Gendervoiceclap_attribute:Gender-0.353-0.20-0.140+0.01-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.932+1.021.072+1.540.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.379-1.231.329-1.361.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.654-0.130.689-0.040.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability1.525+0.871.654+1.111.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.785+0.530.965+1.080.6080.332
162Interestvoiceclap_attribute:Interest1.377-0.501.448-0.341.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.845+0.740.735+0.500.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.955+1.771.000+1.950.5230.245
165Longingvoiceclap_attribute:Longing0.470-0.310.414-0.560.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice1.281+1.301.398+1.630.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive1.966+0.321.960+0.311.7750.594
168Painvoiceclap_attribute:Pain0.441+0.270.644+1.200.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.728+0.750.797+0.980.5030.301
170Pridevoiceclap_attribute:Pride0.925+0.410.892+0.320.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.343+0.092.363+0.152.3130.328
172Reliefvoiceclap_attribute:Relief1.755+1.761.723+1.681.0250.415
173Sadnessvoiceclap_attribute:Sadness0.681-0.230.728-0.070.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous1.327+0.921.452+1.140.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust1.017+0.911.195+1.340.6350.419
176Shamevoiceclap_attribute:Shame0.665+0.490.630+0.370.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh-0.220-1.05-0.265-1.180.1600.360
178Sournessvoiceclap_attribute:Sourness0.814+0.370.845+0.480.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant1.046+1.181.072+1.250.6230.359
180Teasingvoiceclap_attribute:Teasing1.148+1.601.171+1.660.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.537+0.291.700+0.601.3840.529
182Triumphvoiceclap_attribute:Triumph1.244+1.471.348+1.760.7160.359
183Valencevoiceclap_attribute:Valence0.159+0.26-0.083-0.12-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached1.160-0.861.045-1.071.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold-0.253-0.47-0.277-0.53-0.0660.400
186durationvoiceclap_attribute:duration7.479-0.547.726-0.5010.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.429+0.373.467+0.453.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.777+0.824.834+1.214.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.831+0.472.824+0.462.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.887-0.051.882-0.131.8910.069
191talking speedvoiceclap_attribute:talking_speed10.145+0.3010.416+0.379.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.18520.13120.1396
1-0.0014-0.0448-0.0804
20.0180-0.01220.0170
3-0.0297-0.0499-0.0839
40.23150.20740.1769
50.08410.07220.0928
6-0.0615-0.0488-0.0175
70.02860.0154-0.0589
80.01390.03130.0388
9-0.01610.0167-0.0036
100.18520.17150.2001
11-0.0877-0.0626-0.0967
12-0.01000.00520.0208
13-0.0685-0.0906-0.0816
14-0.0352-0.0027-0.0232
15-0.0292-0.0884-0.0741
16-0.1163-0.1610-0.1024
17-0.00480.0103-0.0267
180.01880.04530.0892
190.03060.03850.0414
20-0.0563-0.0363-0.0828
210.15710.20300.1462
22-0.0764-0.0809-0.0778
23-0.0390-0.0630-0.0978
24-0.0327-0.0418-0.0668
25-0.0926-0.0748-0.0726
26-0.0551-0.03630.0296
27-0.0908-0.0783-0.0771
28-0.0816-0.0962-0.0622
290.12000.11200.1599
300.06760.06780.0888
310.10110.12950.1538
320.0075-0.0113-0.0797
330.18150.14700.1891
34-0.0393-0.0691-0.0660
35-0.0536-0.0284-0.0151
36-0.0481-0.0665-0.0149
370.07310.04640.0363
38-0.1029-0.0674-0.0826
39-0.0761-0.1015-0.0834
400.04450.0348-0.0728
41-0.0396-0.02520.0768
42-0.0877-0.0370-0.0834
43-0.0640-0.0523-0.0851
44-0.0393-0.0028-0.0375
45-0.0524-0.0687-0.0513
46-0.0099-0.0195-0.0780
47-0.0822-0.0871-0.0836
48-0.0019-0.0409-0.0531
49-0.1084-0.1164-0.0836
500.21560.20560.1376
510.08590.04590.0349
520.01150.03020.0109
53-0.0938-0.1137-0.0844
54-0.0478-0.0295-0.0634
55-0.0187-0.0194-0.0150
560.01150.03460.0424
57-0.0944-0.0932-0.0907
58-0.1200-0.1505-0.0880
59-0.0393-0.0429-0.0830
60-0.0402-0.0127-0.0341
610.0145-0.0097-0.0358
62-0.0609-0.0809-0.0773
63-0.0643-0.0396-0.0786
640.12790.12770.1487
65-0.0883-0.0827-0.0674
660.03690.03570.0328
670.16080.12950.1340
680.16930.22310.1655
69-0.1011-0.0853-0.0758
700.05090.04420.0925
710.0171-0.0188-0.0396
72-0.0673-0.0518-0.0485
73-0.0387-0.0604-0.0102
74-0.0282-0.0261-0.0377
750.05180.00950.0614
76-0.00900.01090.0503
770.18880.18900.1826
78-0.0047-0.02130.0506
79-0.0962-0.0687-0.0717
800.19130.16270.2051
81-0.0128-0.01510.0494
820.00940.04000.1610
83-0.00910.0064-0.0155
84-0.0612-0.03040.0382
85-0.0388-0.0674-0.0143
860.0021-0.01130.0013
87-0.0215-0.0224-0.0229
880.0100-0.0212-0.0231
890.11820.11460.1146
90-0.0597-0.1019-0.0897
910.0212-0.01050.1216
92-0.0841-0.0831-0.0803
93-0.0346-0.0304-0.0808
94-0.0536-0.03850.0251
95-0.1090-0.1356-0.0889
960.0064-0.0001-0.0209
97-0.1529-0.1164-0.0884
98-0.0968-0.0779-0.0836
990.03960.04350.0250
1000.10230.09270.1678
1010.06520.05140.0966
1020.03610.02490.0268
103-0.0487-0.0536-0.0491
104-0.0822-0.0709-0.0809
105-0.1450-0.1098-0.0798
106-0.0512-0.0444-0.0441
1070.16450.19250.1847
1080.11330.16360.1945
109-0.0834-0.0805-0.0811
1100.15960.15660.1078
111-0.0780-0.0656-0.0959
112-0.1109-0.1120-0.0815
113-0.1632-0.1540-0.1164
114-0.0260-0.0717-0.0126
1150.0051-0.01200.0339
116-0.0804-0.1137-0.0709
1170.00780.01670.0602
1180.0109-0.0168-0.0743
119-0.0003-0.01940.0060
1200.0014-0.02070.0548
121-0.1169-0.1172-0.0838
122-0.1310-0.1207-0.0850
1230.24360.25550.1511
124-0.0245-0.03500.0442
125-0.0767-0.0853-0.0688
126-0.0719-0.0612-0.0713
127-0.0475-0.0166-0.0466
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.0369-0.0756-0.0442
1-0.0758-0.0885-0.0389
2-0.0514-0.1035-0.0494
3-0.0628-0.0756-0.0423
4-0.0446-0.0296-0.0011
50.0253-0.0100-0.0420
60.06880.02520.0119
70.15460.21390.2213
8-0.0137-0.0259-0.0743
9-0.0109-0.0205-0.0181
100.0026-0.0266-0.0421
11-0.0297-0.0310-0.0544
12-0.0698-0.0606-0.0440
13-0.0097-0.0407-0.0355
14-0.0768-0.0681-0.0573
150.03340.05930.0181
160.02980.05380.0646
17-0.0251-0.0271-0.0482
18-0.0002-0.0147-0.0409
19-0.0256-0.0511-0.0422
20-0.0178-0.0368-0.0025
21-0.0633-0.0613-0.0373
22-0.0598-0.0453-0.0417
230.09670.09670.1671
24-0.0299-0.0456-0.0343
25-0.0536-0.0463-0.0718
260.07880.05070.0931
27-0.0474-0.0373-0.0327
28-0.0833-0.0732-0.0435
290.03340.05720.0750
300.03590.06880.0915
31-0.0491-0.0436-0.0505
32-0.1277-0.1110-0.0414
330.05110.05280.0235
340.09820.06270.1093
35-0.01180.00890.0462
360.03190.03630.0905
37-0.0104-0.0395-0.0278
380.0025-0.0352-0.0558
39-0.0601-0.0400-0.0599
40-0.0531-0.0541-0.0465
41-0.0372-0.0528-0.0461
420.08630.11580.1214
430.07080.03660.0920
440.01220.01100.0112
450.06680.02440.0386
46-0.0853-0.0729-0.0591
47-0.0982-0.1294-0.0526
48-0.0479-0.0453-0.0429
49-0.0303-0.0086-0.0251
500.00600.00030.0349
51-0.0147-0.0152-0.0095
52-0.0504-0.0417-0.0396
53-0.0132-0.0112-0.0389
54-0.0868-0.0388-0.0679
550.0266-0.0197-0.0381
56-0.0668-0.0698-0.0553
57-0.00990.02160.0124
58-0.0541-0.0402-0.0386
590.02890.07800.0659
60-0.0621-0.0811-0.0421
610.16560.14030.1627
62-0.0213-0.0358-0.0353
63-0.0688-0.0988-0.0436
640.06380.05590.0776
650.01000.0061-0.0308
660.01800.01090.0804
670.02540.02690.0304
680.08680.04340.0279
690.21640.17100.2118
70-0.0394-0.0400-0.0187
710.01910.00740.0349
72-0.0107-0.0407-0.0445
73-0.0793-0.03010.0151
74-0.0197-0.0247-0.0151
75-0.0596-0.0216-0.0585
760.08830.05350.1129
770.02270.07150.0153
78-0.00410.0332-0.0233
790.04740.05210.0297
800.08880.09470.1225
81-0.0967-0.0773-0.0481
820.0277-0.0122-0.0096
83-0.0559-0.0450-0.0357
840.0347-0.00570.0493
85-0.0818-0.0879-0.0429
86-0.0459-0.0261-0.0383
870.0109-0.0041-0.0191
88-0.1112-0.1035-0.0424
89-0.0536-0.0220-0.0388
900.0041-0.0026-0.0312
910.14860.16690.0573
92-0.0237-0.0156-0.0371
93-0.0186-0.0221-0.0487
94-0.0838-0.0685-0.0695
950.0384-0.01430.0224
96-0.0823-0.0691-0.0384
97-0.0564-0.0848-0.0505
980.22340.20710.2803
99-0.0564-0.0606-0.0486
1000.10920.12120.0637
101-0.0072-0.0257-0.0405
1020.0202-0.0366-0.0479
103-0.0068-0.0158-0.0690
104-0.0466-0.0712-0.0559
105-0.1297-0.1103-0.0558
1060.03120.01680.0766
1070.05090.03410.1396
108-0.0633-0.0582-0.0319
1090.00360.0140-0.0057
1100.04060.01730.0213
1110.07630.06810.0835
112-0.0142-0.0238-0.0346
1130.04440.04630.0601
1140.05560.04630.0796
1150.02180.0361-0.0304
116-0.00160.0158-0.0003
1170.05640.03900.0376
118-0.0628-0.0130-0.0560
1190.06010.03510.1198
120-0.0556-0.0654-0.0214
1210.07380.07250.0021
122-0.00140.0201-0.0228
123-0.05930.0018-0.0443
124-0.0516-0.0742-0.0428
1250.0231-0.0346-0.0417
1260.09180.07870.0502
127-0.0207-0.0080-0.0179
128-0.0231-0.0344-0.0500
129-0.0504-0.0429-0.0455
130-0.0499-0.0742-0.0431
131-0.0424-0.0129-0.0375
1320.08880.04730.0565
1330.01030.01910.1025
134-0.0606-0.01080.0217
1350.09530.02320.0538
136-0.0561-0.0606-0.0460
137-0.0156-0.0150-0.0398
138-0.0673-0.0548-0.0378
1390.06680.09940.1250
140-0.0159-0.0434-0.0441
141-0.0041-0.0255-0.0038
142-0.0227-0.0456-0.0371
143-0.1122-0.0967-0.0803
1440.0399-0.0198-0.0447
145-0.0618-0.0538-0.0545
1460.13560.13350.1177
1470.03270.03390.0640
148-0.0601-0.0623-0.0371
149-0.0653-0.0125-0.0552
1500.02730.03930.0771
1510.02670.07220.0739
152-0.0132-0.0320-0.0139
1530.05910.01810.0235
1540.05260.06640.0697
155-0.0043-0.0068-0.0408
156-0.0501-0.0427-0.0430
1570.02620.04090.0013
158-0.0798-0.0548-0.0534
159-0.02070.0276-0.0474
1600.04860.03760.0473
161-0.1177-0.0954-0.0939
1620.04890.00840.0023
1630.00320.02610.0317
1640.03190.05380.0764
165-0.0196-0.0181-0.0614
166-0.0933-0.0940-0.0499
167-0.01020.01790.0094
1680.08580.09740.0916
169-0.0441-0.0296-0.0489
170-0.0491-0.0518-0.0402
171-0.0327-0.0235-0.0377
1720.0055-0.0220-0.0271
173-0.0967-0.0633-0.0092
1740.07580.07530.1748
175-0.0494-0.0470-0.0113
176-0.02290.0131-0.0365
177-0.0451-0.0507-0.0353
1780.02170.05760.0279
179-0.0298-0.0606-0.0528
1800.13560.14100.0528
181-0.0449-0.0325-0.0574
182-0.0386-0.0524-0.0370
183-0.0188-0.0169-0.0158
184-0.0342-0.0375-0.0435
185-0.04010.0250-0.0513
186-0.0529-0.0298-0.0517
187-0.02510.0254-0.0488
188-0.0266-0.0378-0.0068
189-0.04760.0130-0.0045
1900.0058-0.0189-0.0272
1910.05810.05720.0598
192-0.0499-0.0596-0.0429
1930.08730.04100.0684
194-0.0138-0.0267-0.0556
1950.04810.07870.0743
196-0.0016-0.02670.0383
1970.01880.05070.0615
198-0.0034-0.01760.1100
1990.08330.17370.1192
2000.01340.0257-0.0220
2010.09230.12870.0759
202-0.0933-0.0800-0.0359
203-0.0718-0.0354-0.0535
2040.07430.08310.1078
2050.02890.04310.0332
206-0.0858-0.0845-0.0442
2070.00840.01820.0690
208-0.0858-0.0879-0.0445
2090.0132-0.01920.0136
210-0.0119-0.0358-0.0558
211-0.0153-0.0189-0.0330
2120.01060.0240-0.0627
2130.07830.11310.1367
2140.07730.14100.0812
2150.03020.04390.0482
2160.06260.04240.0665
2170.05340.03100.0602
218-0.0429-0.0688-0.0498
219-0.0342-0.0346-0.0558
220-0.0065-0.02110.0676
221-0.1172-0.0865-0.0372
222-0.1262-0.1103-0.0503
2230.03640.0139-0.0021
2240.06780.03410.1515
225-0.0596-0.0364-0.0357
2260.08530.14240.1075
227-0.0206-0.02720.0024
228-0.0554-0.0363-0.0185
229-0.0873-0.0933-0.0367
2300.0117-0.0272-0.0271
231-0.01010.0060-0.0212
232-0.0718-0.0620-0.0589
2330.12470.08340.0635
234-0.0496-0.0514-0.0364
2350.19450.22610.1080
236-0.0303-0.0335-0.0099
2370.03620.02790.0092
238-0.0733-0.0610-0.0470
2390.0446-0.0228-0.0112
240-0.0344-0.0109-0.0437
241-0.0668-0.0681-0.0579
242-0.1232-0.1131-0.0632
2430.06830.02720.0181
2440.11070.07460.0627
2450.02280.0366-0.0408
2460.01630.0257-0.0347
247-0.0070-0.0210-0.0361
248-0.0145-0.0278-0.0231
249-0.0216-0.0647-0.0654
Source: P3 source-disjoint test split · p3_02005463 · SHA-256 d9d7139aaf95b3cd…
p04Speech + burstsP3 curated burst test

Fear + 4 known burst(s)

Synthetic speech with inserted bursts · P3 synthetic speech/burst mixture · 8.0 s

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Astonishment Surprise
z +3.80 / +3.80raw 2.18 / 2.18
Disgust
z +1.29 / +2.91raw 0.64 / 1.43
Fear
z +2.20 / +2.35raw 1.35 / 1.45
Impatience and Irritability
z +2.32 / +1.84raw 1.79 / 1.44
Contempt
z +1.31 / +2.08raw 0.61 / 0.97
Anger
z +2.05 / +2.02raw 1.26 / 1.24
Teasing
z +1.25 / +1.71raw 0.68 / 0.93

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Pitch range (RANG)
z +4.10 / +4.07raw 6.59 / 6.56
Volatility (VOLT)
z +3.80 / +3.39raw 5.95 / 5.59
Dramatic style (S_DRAM)
z +3.45 / +3.27raw 6.69 / 6.47
Dynamic arc (DARC)
z +3.31 / +3.37raw 6.19 / 6.27
Roughness (ROUG)
z +2.89 / +3.01raw 4.59 / 4.70
Ranting or angry style (S_RANT)
z +2.83 / +2.27raw 4.86 / 4.15
Tension (TENS)
z +2.65 / +2.79raw 3.82 / 3.94

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z -0.14 / -0.32raw 2.10 / 1.81
Vocal-burst blend (0–10)z -1.84 / -1.78raw 1.63 / 1.80
Content enjoyment (CE)z -0.73 / -0.94raw 4.05 / 3.85
Production quality (PQ)z +0.31 / +0.08raw 7.31 / 7.11
Speech signal quality, calibratedz -0.23 / -0.14raw 3.37 / 3.42
Background quality, calibratedz -1.44 / -1.16raw 2.05 / 2.31
Overall quality, calibratedz -1.18 / -0.91raw 2.13 / 2.29
P.808 MOS estimatez -0.97 / -0.96raw 2.94 / 2.95
score content enjoymentz -1.02 / -0.63raw 4.20 / 4.32
score overall qualityz -0.47 / -0.44raw 2.41 / 2.43
Background Noisez +0.84 / +0.45raw 1.03 / 0.93
Tempo (TEMP)z +1.35 / +1.27raw 3.72 / 3.63
Arousal (AROU)z +2.31 / +2.51raw 5.01 / 5.24
Valence (VALN)z -0.71 / -0.08raw 1.53 / 2.23
Recording quality (RCQL)z -0.85 / -0.81raw 2.00 / 2.03
Background noise (BKGN)z -0.31 / -0.18raw 2.47 / 2.56
talking speedz +0.14 / -0.00raw 9.56 / 9.01
Characters per second (CPS): 8.57 raw · z -0.12

P3 metadata teacher estimate; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.7740.790
Identity · 250D0.7640.786

Vocal-burst regions

Known
Tiny
Base
0 s8.0 s

Purple marks known inserted burst intervals; other vocal sounds may be unlabeled.

Tiny: 3 regions, count-head estimate 3.82; Base: 5 regions, count-head estimate 3.75.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
Tiny10.300.900.60
Tiny21.005.664.66
Tiny35.706.100.40
Base10.300.900.60
Base21.001.200.20
Base31.241.260.02
Base41.305.664.36
Base55.706.100.40

Known inserted intervals: 0.31–5.33 s, 5.33–5.64 s, 5.70–6.10 s, 0.38–0.89 s

All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection-0.064-0.13-0.020-0.050.0070.570
1Amusementemo_Amusement0.910+1.031.057+1.20-0.0010.885
2Angeremo_Anger1.263+2.051.244+2.020.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise2.176+3.802.176+3.800.0050.572
4Aweemo_Awe0.027+0.060.071+0.170.0000.427
5Bitternessemo_Bitterness0.177+0.390.178+0.400.0130.415
6Concentrationemo_Concentration0.257-0.510.240-0.530.6340.741
7Confusionemo_Confusion0.778+1.440.743+1.370.0010.540
8Contemplationemo_Contemplation0.027-0.090.034-0.080.0820.598
9Contemptemo_Contempt0.609+1.310.967+2.08-0.0000.464
10Contentmentemo_Contentment-0.094-0.18-0.041-0.080.0000.514
11Disappointmentemo_Disappointment0.473+0.680.008+0.010.0000.700
12Disgustemo_Disgust0.636+1.291.433+2.910.0070.489
13Distressemo_Distress1.334+1.621.110+1.350.0000.821
14Doubtemo_Doubt0.166+0.340.174+0.360.0060.466
15Elationemo_Elation0.250+0.330.192+0.250.0120.728
16Embarrassmentemo_Embarrassment0.158+0.270.201+0.370.0430.430
17Emotional Numbnessemo_Emotional_Numbness-0.203-0.230.075+0.080.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.038+0.020.042+0.030.0280.539
19Fearemo_Fear1.354+2.201.450+2.350.0150.610
20Helplessnessemo_Helplessness0.731+1.060.409+0.59-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.030-0.120.089-0.040.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability1.786+2.321.444+1.840.1490.705
23Infatuationemo_Infatuation-0.011+0.010.032+0.09-0.0150.533
24Interestemo_Interest1.071-0.571.234-0.391.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness0.175+0.360.112+0.24-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.086+0.060.019-0.090.0590.442
27Longingemo_Longing0.043-0.070.028-0.090.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.609+1.040.877+1.540.0550.534
29Painemo_Pain0.697+1.210.447+0.77-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.249+0.380.293+0.450.0170.614
31Prideemo_Pride-0.052-0.11-0.039-0.080.0020.509
32Reliefemo_Relief0.106+0.110.144+0.160.0330.681
33Sadnessemo_Sadness0.374+0.490.110+0.140.0000.759
34Sexual Lustemo_Sexual_Lust0.260+0.470.415+0.780.0300.494
35Shameemo_Shame0.031+0.040.018+0.010.0130.510
36Sournessemo_Sourness0.287+0.670.306+0.71-0.0050.439
37Teasingemo_Teasing0.680+1.250.930+1.710.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.014-0.040.045+0.010.0410.697
39Triumphemo_Triumph0.180+0.360.017+0.05-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg0.929-1.661.205-1.442.9841.238
41Arousal (AROU)vn_AROU_reg5.007+2.315.235+2.512.3591.147
42Arousal shift (ARSH)vn_ARSH_reg3.957+1.594.238+1.912.5800.867
43Vocal attack (ATCK)vn_ATCK_reg3.919+1.364.330+1.672.1621.295
44Background noise (BKGN)vn_BKGN_reg2.467-0.312.564-0.182.7030.761
45Brightness (BRGT)vn_BRGT_reg2.768+0.963.129+1.401.9910.813
46Speech chunking (CHNK)vn_CHNK_reg2.335+0.392.447+0.481.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg1.166-0.901.448-0.602.0180.945
48Cognitive load (COGL)vn_COGL_reg1.120-0.920.868-1.162.0941.061
49Dynamic arc (DARC)vn_DARC_reg6.189+3.316.274+3.371.6151.380
50Disfluency (DFLU)vn_DFLU_reg3.355+0.732.718+0.192.4881.184
51Emphasis (EMPH)vn_EMPH_reg4.522+1.264.596+1.302.6071.524
52Esthetics (ESTH)vn_ESTH_reg1.212-1.521.467-1.202.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg-0.001-0.27-0.017-0.320.0860.322
54Focus (FOCS)vn_FOCS_reg1.656-0.621.925-0.382.3441.102
55Fullness (FULL)vn_FULL_reg0.860-1.031.147-0.711.7780.889
56Perceived gender (GEND)vn_GEND_reg2.084-1.242.715-0.844.0551.588
57Harmonicity (HARM)vn_HARM_reg2.648+0.202.857+0.452.4800.836
58Metallic character (METL)vn_METL_reg3.704+2.174.108+2.601.6210.958
59Pitch range (RANG)vn_RANG_reg6.590+4.106.562+4.072.9140.897
60Recording quality (RCQL)vn_RCQL_reg1.996-0.852.031-0.812.7540.888
61Vocal register (REGS)vn_REGS_reg3.562+1.863.631+1.921.1981.268
62Respiration (RESP)vn_RESP_reg4.660+1.884.309+1.542.6801.056
63Roughness (ROUG)vn_ROUG_reg4.592+2.894.702+3.012.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg0.721-1.670.987-1.362.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg3.405+1.533.250+1.342.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg2.898+0.983.202+1.291.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg0.003-1.700.513-1.101.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg4.219+1.364.384+1.522.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg3.155+1.313.254+1.442.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg2.230+0.202.486+0.522.0700.803
71Smoothness (SMTH)vn_SMTH_reg1.043-1.011.069-0.982.1431.093
72Stance (STNC)vn_STNC_reg4.335+1.504.291+1.462.6331.137
73Speech structure (STRU)vn_STRU_reg0.997-1.191.112-1.072.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg0.418-0.670.429-0.671.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg0.246-0.810.436-0.621.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg5.359+2.715.423+2.761.5601.401
77Casual style (S_CASU)vn_S_CASU_reg5.256+1.144.887+0.863.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg2.475-0.092.212-0.292.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg6.687+3.456.473+3.272.7211.148
80Formal style (S_FORM)vn_S_FORM_reg0.080-1.090.261-0.921.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg1.215-1.251.265-1.212.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg0.522-0.540.799-0.351.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.124-0.500.137-0.480.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg3.487+0.863.721+1.042.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg4.863+2.834.146+2.271.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg4.988+1.834.977+1.832.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg0.103-0.440.072-0.470.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg0.886-0.440.893-0.441.5971.618
89Tempo (TEMP)vn_TEMP_reg3.719+1.353.635+1.272.1171.190
90Tension (TENS)vn_TENS_reg3.818+2.653.935+2.791.5590.852
91Valence (VALN)vn_VALN_reg1.533-0.712.234-0.082.3281.118
92Valence shift (VALS)vn_VALS_reg2.313-0.192.423-0.092.5181.079
93Velocity flux (VFLX)vn_VFLX_reg5.064+2.324.913+2.202.2561.208
94Volatility (VOLT)vn_VOLT_reg5.955+3.805.588+3.392.5860.886
95Vulnerability (VULN)vn_VULN_reg4.466+1.854.257+1.632.6480.984
96Warmth (WARM)vn_WARM_reg1.347-1.351.467-1.222.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_62.097-0.141.815-0.322.3321.621
98Vocal-burst blend (0–10)blend_0_101.633-1.841.803-1.786.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.542-1.640.513-1.721.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age2.761-1.672.986-0.903.2500.293
101Arousaleiv_extra:Arousal2.894+3.692.765+3.391.2890.435
102Authenticityeiv_extra:Authenticity2.825+1.042.796+0.842.6670.153
103Background Noiseeiv_extra:Background_Noise1.029+0.840.928+0.450.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.660+2.051.454+1.371.0400.303
105Gendereiv_extra:Gender0.596+1.220.616+1.24-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.399-1.711.427-1.601.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive2.701+1.792.583+1.541.8590.470
108Recording Qualityeiv_extra:Recording_Quality1.522-1.871.707-1.322.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous1.302+1.091.421+1.440.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.338+0.180.338+0.180.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.394+0.060.566+0.590.3760.324
112Valenceeiv_extra:Valence-0.355-0.95-0.281-0.750.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached0.870-1.931.064-1.491.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold-0.221-0.91-0.200-0.840.0510.300
115score background qualityeiv_extra:score_background_quality2.953-0.502.932-0.533.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.198-1.024.316-0.634.5070.303
117score overall qualityeiv_extra:score_overall_quality2.415-0.472.427-0.442.6520.507
118score speech qualityeiv_extra:score_speech_quality1.801-0.051.802-0.041.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE4.053-0.733.846-0.944.7991.018
120Content usefulness (CU)audiobox:CU5.413-0.245.285-0.365.6491.001
121Production complexity (PC)audiobox:PC2.647+0.202.262-0.042.3311.617
122Production quality (PQ)audiobox:PQ7.313+0.317.110+0.087.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw3.707-0.293.785-0.173.9000.669
124Background quality, rawdnsmos:BAK_raw1.634-1.401.977-1.103.2111.124
125Overall quality, rawdnsmos:OVRL_raw2.112-1.202.351-0.903.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.373-0.233.416-0.143.4880.501
127Background quality, calibrateddnsmos:BAK2.051-1.442.312-1.163.3890.928
128Overall quality, calibrateddnsmos:OVRL2.127-1.182.288-0.912.8400.606
129P.808 MOS estimatednsmos:P808_MOS2.941-0.972.945-0.963.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p1.573+2.431.558+2.410.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection1.245+0.541.242+0.530.9470.554
132Agevoiceclap_attribute:Age1.841-2.541.960-2.323.2290.547
133Amusementvoiceclap_attribute:Amusement1.940+2.712.029+2.860.3920.572
134Angervoiceclap_attribute:Anger1.390+0.791.227+0.460.9920.506
135Arousalvoiceclap_attribute:Arousal3.060+2.642.963+2.481.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise1.609+2.291.467+2.000.5090.479
137Authenticityvoiceclap_attribute:Authenticity3.069+1.793.013+1.452.7700.167
138Awevoiceclap_attribute:Awe0.639+0.390.636+0.380.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.845+1.050.751+0.700.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.803+0.890.766+0.740.5920.236
141Concentrationvoiceclap_attribute:Concentration0.149-2.010.137-2.031.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.785+1.531.654+1.251.0690.468
143Confusionvoiceclap_attribute:Confusion0.941+0.890.939+0.890.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.088-1.710.176-1.490.7800.406
145Contemptvoiceclap_attribute:Contempt0.668-0.120.701-0.020.7090.339
146Contentmentvoiceclap_attribute:Contentment0.702+0.370.720+0.430.5880.307
147Disappointmentvoiceclap_attribute:Disappointment1.165+1.391.085+1.120.7600.291
148Disgustvoiceclap_attribute:Disgust0.731+0.510.661+0.180.6250.207
149Distressvoiceclap_attribute:Distress1.768+2.411.585+1.880.9270.350
150Doubtvoiceclap_attribute:Doubt0.195-1.810.203-1.760.5350.188
151Elationvoiceclap_attribute:Elation1.751+3.051.701+2.930.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.876+1.810.724+1.330.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.165-1.880.231-1.630.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.617+0.590.500+0.150.4620.261
155Fearvoiceclap_attribute:Fear0.977+1.690.891+1.290.6180.212
156Gendervoiceclap_attribute:Gender1.034+1.180.862+1.01-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness1.281+2.321.017+1.340.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.248-1.581.329-1.361.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism1.553+2.141.500+2.010.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability1.705+1.211.419+0.681.0520.541
161Infatuationvoiceclap_attribute:Infatuation1.274+2.011.155+1.650.6080.332
162Interestvoiceclap_attribute:Interest1.243-0.811.398-0.461.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness1.023+1.120.928+0.920.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.908+1.580.846+1.320.5230.245
165Longingvoiceclap_attribute:Longing0.773+1.070.734+0.890.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice1.160+0.951.106+0.790.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive3.093+2.222.977+2.021.7750.594
168Painvoiceclap_attribute:Pain1.105+3.320.894+2.350.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy1.570+3.551.575+3.560.5030.301
170Pridevoiceclap_attribute:Pride0.802+0.060.888+0.310.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality1.807-1.541.825-1.492.3130.328
172Reliefvoiceclap_attribute:Relief1.469+1.071.185+0.391.0250.415
173Sadnessvoiceclap_attribute:Sadness1.361+1.951.193+1.410.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous1.980+2.071.989+2.090.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust1.407+1.841.300+1.580.6350.419
176Shamevoiceclap_attribute:Shame0.449-0.270.292-0.830.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.401+0.670.485+0.900.1600.360
178Sournessvoiceclap_attribute:Sourness0.682-0.080.639-0.230.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.640+0.050.557-0.190.6230.359
180Teasingvoiceclap_attribute:Teasing1.222+1.801.391+2.240.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.523+0.261.453+0.131.3840.529
182Triumphvoiceclap_attribute:Triumph1.121+1.131.216+1.390.7160.359
183Valencevoiceclap_attribute:Valence0.053+0.090.442+0.71-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached0.576-1.910.767-1.571.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold0.150+0.540.402+1.17-0.0660.400
186durationvoiceclap_attribute:duration5.623-0.835.821-0.8010.9146.354
187score background qualityvoiceclap_attribute:score_background_quality2.915-0.662.876-0.743.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.591-0.444.634-0.154.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.295-0.832.249-0.942.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.914+0.341.912+0.301.8910.069
191talking speedvoiceclap_attribute:talking_speed9.563+0.149.014-0.009.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.06920.11540.1144
10.08180.0603-0.0244
2-0.0139-0.0329-0.0472
3-0.02270.02320.1272
4-0.1004-0.1080-0.0646
50.00730.0088-0.0532
6-0.0827-0.0932-0.0623
70.13750.12950.0903
80.11640.06320.0607
90.0028-0.00120.0287
10-0.1034-0.1036-0.0639
110.03000.08100.0527
12-0.1181-0.0914-0.0691
130.08480.10360.1885
14-0.0010-0.0038-0.0556
150.12910.17830.1720
160.05530.0533-0.0057
170.0196-0.0020-0.0715
18-0.1042-0.0906-0.0634
19-0.0167-0.0141-0.0650
200.0050-0.0314-0.0688
210.07000.1095-0.0155
22-0.0443-0.0122-0.0646
230.20330.24560.0922
240.10340.11610.0870
250.14430.07430.1592
26-0.0949-0.1043-0.0645
270.05650.0274-0.0214
28-0.1059-0.0984-0.0757
29-0.0340-0.0740-0.0736
300.07590.04770.0531
31-0.0886-0.0939-0.0258
320.00140.0359-0.0097
330.06540.06100.0017
340.01840.02530.0795
35-0.0489-0.0108-0.0529
36-0.0844-0.0729-0.0739
370.08180.0928-0.0590
380.15190.15240.2252
390.12570.12060.0073
400.05130.07030.0994
410.03630.0140-0.0988
42-0.0776-0.1013-0.0693
430.01420.01850.0207
440.15780.16420.1545
450.06120.02640.1210
46-0.0974-0.0847-0.0648
47-0.0426-0.03880.0036
48-0.1097-0.1110-0.0671
49-0.0759-0.0566-0.0480
500.01530.0429-0.0703
51-0.0591-0.01850.0611
520.11470.14420.0897
530.03100.0261-0.0633
54-0.0751-0.0673-0.0769
55-0.0382-0.06990.0164
56-0.0953-0.0999-0.0652
570.0248-0.00080.0564
58-0.0205-0.0385-0.0374
59-0.0084-0.0396-0.0392
600.08480.12870.1082
61-0.0432-0.0485-0.0633
620.07800.10580.0597
630.10670.13610.1909
640.02590.04110.0632
65-0.0882-0.0743-0.0561
660.0293-0.00350.1693
67-0.0705-0.0327-0.0663
68-0.0454-0.0435-0.0726
690.02370.0235-0.0630
700.07890.05140.1284
71-0.0797-0.0621-0.0762
720.06710.04960.0373
73-0.1008-0.0895-0.0840
740.05530.05810.0398
750.04680.0492-0.0028
760.04300.02810.1004
77-0.0802-0.0925-0.0498
78-0.0886-0.0976-0.0612
790.12820.0969-0.0048
80-0.1131-0.1013-0.0622
810.05610.05580.0958
82-0.0700-0.0533-0.0657
83-0.1198-0.1213-0.0704
840.07760.04480.0159
85-0.0759-0.0566-0.0730
860.13670.15760.1644
870.05780.03090.1038
88-0.1059-0.0954-0.0638
89-0.1156-0.0954-0.0630
90-0.0130-0.0255-0.0251
91-0.0751-0.0681-0.0632
92-0.0751-0.0817-0.0667
930.26320.25000.2118
940.05150.07770.1335
950.02400.0392-0.0068
960.03750.01670.1440
970.12740.10430.1304
980.08520.08060.0999
990.20840.18200.0998
100-0.1190-0.1228-0.0729
101-0.1181-0.0747-0.0665
102-0.0650-0.0221-0.0206
103-0.0188-0.0158-0.0456
104-0.0008-0.0129-0.0518
105-0.0894-0.0851-0.0663
106-0.1434-0.1243-0.0826
107-0.1181-0.1147-0.0640
108-0.1282-0.1258-0.0745
1090.05360.05140.1267
1100.03180.06730.1985
1110.09620.09250.1438
112-0.0067-0.0212-0.0169
113-0.0882-0.0714-0.0849
114-0.0321-0.0677-0.0740
1150.16870.23520.1576
1160.0460-0.01120.0470
1170.00510.00570.0944
118-0.1025-0.1065-0.0669
1190.0098-0.0388-0.0665
1200.11310.08510.0212
1210.17300.17160.1913
122-0.0903-0.0847-0.0706
1230.14090.11540.1121
124-0.0548-0.0366-0.0007
1250.08560.06070.0323
1260.05060.07770.1240
1270.03020.01740.0541
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.0446-0.03630.0301
1-0.0749-0.0584-0.0362
2-0.0876-0.0655-0.0460
3-0.01970.0157-0.0105
40.0040-0.01250.0355
5-0.0886-0.0821-0.0391
60.09990.10350.1417
7-0.0539-0.0314-0.0379
8-0.0399-0.0196-0.0692
90.09700.08650.1494
10-0.0040-0.00950.0673
11-0.0122-0.0500-0.0448
120.07000.0782-0.0056
130.07540.09400.1388
14-0.0289-0.0655-0.0534
15-0.0544-0.0632-0.0441
160.02800.0134-0.0211
17-0.0455-0.0401-0.0450
18-0.0086-0.0391-0.0381
19-0.0098-0.01690.0085
200.02280.09080.1285
21-0.0468-0.0470-0.0347
22-0.0566-0.0628-0.0388
230.00580.0391-0.0362
240.0214-0.01170.0096
250.05530.06240.0799
260.02450.08330.0756
27-0.0081-0.0065-0.0304
28-0.0326-0.0255-0.0405
29-0.0146-0.0332-0.0360
300.0252-0.00930.0399
31-0.0671-0.0817-0.0470
320.0145-0.0663-0.0385
33-0.03530.0042-0.0336
340.08960.0529-0.0175
35-0.0521-0.0727-0.0131
360.06510.10740.0167
370.01640.00500.0373
38-0.02550.0151-0.0385
390.06900.0588-0.0222
40-0.0428-0.0608-0.0434
41-0.0563-0.0667-0.0430
420.04480.02820.0057
430.06240.0027-0.0443
440.02190.0196-0.0193
450.07050.08570.0891
46-0.0749-0.0817-0.0550
470.05560.03610.0819
48-0.0002-0.0300-0.0196
49-0.0372-0.0865-0.0335
50-0.0335-0.0389-0.0198
510.14690.16660.1975
52-0.0588-0.0719-0.0369
53-0.0592-0.0332-0.0362
540.0436-0.00070.0147
55-0.00370.0118-0.0353
56-0.0769-0.0758-0.0515
57-0.0641-0.0727-0.0437
58-0.0857-0.0569-0.0359
590.03130.01650.0770
60-0.0106-0.0312-0.0392
610.09990.08530.1299
62-0.0079-0.0078-0.0329
63-0.04090.0062-0.0187
64-0.0083-0.0624-0.0417
650.00610.0204-0.0032
66-0.1116-0.0496-0.0472
670.12140.10350.1042
680.0081-0.01300.0123
690.02860.03470.0484
700.0076-0.01380.0355
71-0.0590-0.0270-0.0392
720.04600.0703-0.0341
730.09060.02700.0337
740.02090.06520.0609
75-0.0135-0.0436-0.0545
760.08280.10820.0145
770.11310.06160.0087
780.0058-0.0399-0.0085
79-0.02820.0041-0.0400
80-0.0116-0.0048-0.0145
81-0.0700-0.0569-0.0448
82-0.0284-0.0123-0.0345
830.14100.19660.1512
84-0.0779-0.0513-0.0563
85-0.0193-0.0347-0.0400
86-0.0052-0.0403-0.0357
870.0095-0.0180-0.0319
880.05090.04500.0207
89-0.0185-0.0369-0.0218
900.03530.0084-0.0396
910.0289-0.0002-0.0329
920.10870.12160.0283
930.02510.01340.0136
94-0.0278-0.01490.0281
95-0.0228-0.0149-0.0417
96-0.0686-0.0616-0.0357
97-0.0710-0.0612-0.0471
980.11510.11850.1099
990.04280.04580.0506
1000.14000.15320.1270
1010.09840.16740.1382
1020.08180.05570.0389
1030.0173-0.0027-0.0660
104-0.0671-0.0496-0.0521
105-0.0734-0.0569-0.0520
1060.13510.11060.3135
1070.05410.03400.0519
1080.03500.0326-0.0246
109-0.0627-0.0561-0.0384
1100.02000.04360.0352
111-0.0049-0.0464-0.0329
112-0.0083-0.03240.0515
1130.06000.06910.0124
114-0.0244-0.0430-0.0557
1150.11560.04860.0201
1160.11310.08170.0530
117-0.0127-0.0316-0.0497
1180.05240.06120.0668
1190.14100.14140.1838
120-0.0798-0.0470-0.0509
1210.12540.15480.0831
1220.02350.01380.0450
1230.0015-0.03670.0129
1240.11260.12080.1417
125-0.0198-0.0624-0.0388
1260.0258-0.00010.0194
1270.05510.09630.1552
128-0.0038-0.0381-0.0209
1290.08420.04380.0533
130-0.0294-0.0320-0.0402
1310.09350.10980.0626
132-0.01220.0185-0.0286
1330.00220.0082-0.0058
1340.05210.08570.0714
1350.0017-0.00870.0127
136-0.0404-0.0231-0.0428
137-0.01190.03610.0518
138-0.0541-0.0509-0.0352
1390.06710.0136-0.0409
1400.04330.01860.0572
1410.07440.04520.0526
1420.02150.02940.0005
143-0.0774-0.0505-0.0309
1440.07250.12320.0983
145-0.0416-0.0149-0.0508
1460.12340.15160.1549
1470.01050.0055-0.0355
148-0.0521-0.0577-0.0441
1490.09450.0857-0.0515
150-0.0397-0.0434-0.0393
151-0.0348-0.0322-0.0455
152-0.0416-0.0466-0.0443
1530.0002-0.0300-0.0625
1540.19980.21950.1510
155-0.0521-0.0561-0.0522
1560.05530.02340.0126
1570.03430.04170.0291
158-0.0162-0.0367-0.0498
159-0.02500.03260.0204
1600.04580.0774-0.0380
1610.02250.0009-0.0314
1620.06810.07500.0497
1630.03620.05770.0942
1640.01980.0397-0.0351
1650.0365-0.0153-0.0045
166-0.0798-0.0655-0.0465
167-0.0553-0.0553-0.0507
1680.01790.01720.0271
169-0.0088-0.0486-0.0456
170-0.0502-0.0347-0.0125
1710.04530.0525-0.0168
1720.00790.02110.0674
173-0.0185-0.0351-0.0318
1740.09840.06200.0379
1750.06190.0500-0.0213
1760.0135-0.0171-0.0340
177-0.01500.01370.0654
1780.07930.10270.0191
179-0.0867-0.0395-0.0492
1800.08670.09990.1085
1810.07300.03870.0453
182-0.0290-0.0391-0.0345
183-0.00270.0166-0.0260
184-0.0588-0.0695-0.0406
185-0.0345-0.0436-0.0478
1860.0242-0.00150.0052
187-0.0588-0.0750-0.0417
188-0.0725-0.0399-0.0375
1890.04190.04150.0791
1900.03620.0513-0.0076
191-0.0160-0.0075-0.0534
192-0.0695-0.0482-0.0484
193-0.0411-0.0632-0.0421
194-0.0499-0.03930.0199
195-0.0283-0.0580-0.0380
196-0.0695-0.0758-0.0388
1970.03430.06200.0673
198-0.0705-0.0754-0.0386
1990.30550.21480.2273
200-0.0282-0.0234-0.0295
2010.02420.0351-0.0372
202-0.0710-0.0809-0.0831
203-0.0783-0.0671-0.0498
204-0.0632-0.0274-0.0330
205-0.06000.0203-0.0090
206-0.0186-0.0063-0.0412
207-0.0867-0.0636-0.0540
208-0.0744-0.0620-0.0414
209-0.0406-0.04460.0341
210-0.02080.0136-0.0475
2110.01220.01470.0626
212-0.0509-0.0369-0.0584
2130.04750.05800.0377
2140.02340.0194-0.0258
215-0.0433-0.0458-0.0336
2160.03750.05730.0353
2170.02780.01970.0181
218-0.0421-0.0636-0.0464
2190.02340.0187-0.0107
2200.02220.00950.0116
2210.08860.08130.1188
2220.06560.04320.0515
2230.16350.12080.1177
224-0.0666-0.0387-0.0365
2250.0268-0.0417-0.0332
2260.0300-0.00110.0110
227-0.0219-0.0185-0.0226
228-0.01850.0108-0.0475
229-0.0921-0.0644-0.0438
230-0.0583-0.0774-0.0407
231-0.0073-0.00920.0724
232-0.0392-0.0379-0.0548
233-0.00910.0318-0.0253
234-0.0612-0.0790-0.0339
2350.0487-0.0167-0.0328
236-0.0155-0.0157-0.0343
237-0.0377-0.0024-0.0357
238-0.0690-0.0604-0.0438
2390.02830.07270.0003
2400.10380.15790.0521
241-0.0590-0.0620-0.0372
242-0.0813-0.0766-0.0588
2430.01690.0557-0.0291
2440.02220.0381-0.0425
245-0.0039-0.0569-0.0380
2460.07830.09670.1232
2470.13320.09950.2095
2480.08520.05050.0285
249-0.0573-0.0391-0.0565
Source: P3 source-disjoint test split · p3_02007567 · SHA-256 a5461e9ad194504a…
p05Speech + burstsP3 curated burst test

Sadness + 1 known burst(s)

Synthetic speech with inserted bursts · P3 synthetic speech/burst mixture · 12.0 s

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Contemplation
z +3.19 / +3.60raw 1.99 / 2.24
Sadness
z +3.31 / +3.42raw 2.51 / 2.60
Confusion
z +2.76 / +3.01raw 1.49 / 1.63
Longing
z +2.80 / +2.45raw 1.86 / 1.64
Doubt
z +2.54 / +2.63raw 1.19 / 1.23
Helplessness
z +2.56 / +2.33raw 1.78 / 1.62
Disappointment
z +2.39 / +2.44raw 1.67 / 1.71

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

ASMR style (S_ASMR)
z +2.47 / +2.59raw 4.85 / 5.02
Whispered talking style (S_WHIS)
z +2.33 / +2.55raw 5.36 / 5.71
Vocal register (REGS)
z +1.96 / +2.06raw 3.68 / 3.82
Narrator style (S_NARR)
z +1.85 / +1.85raw 3.98 / 3.98
Background noise (BKGN)
z +1.56 / +1.33raw 3.89 / 3.72
Newsreader style (S_NEWS)
z +1.48 / +1.51raw 1.25 / 1.27
Articulation clarity (CLRT)
z +1.46 / +1.47raw 3.40 / 3.40

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z -0.85 / -0.86raw 0.96 / 0.93
Vocal-burst blend (0–10)z -2.41 / -2.38raw 0.08 / 0.17
Content enjoyment (CE)z +0.64 / +0.64raw 5.45 / 5.45
Production quality (PQ)z -0.01 / +0.19raw 7.04 / 7.21
Speech signal quality, calibratedz +0.17 / +0.29raw 3.57 / 3.63
Background quality, calibratedz +0.10 / +0.19raw 3.48 / 3.56
Overall quality, calibratedz +0.20 / +0.39raw 2.96 / 3.08
P.808 MOS estimatez +0.38 / +0.43raw 3.61 / 3.63
score content enjoymentz +0.77 / +0.70raw 4.74 / 4.72
score overall qualityz +0.50 / +0.14raw 2.91 / 2.72
Background Noisez -1.24 / -2.50raw 0.48 / 0.15
Tempo (TEMP)z -0.93 / -0.89raw 1.01 / 1.05
Arousal (AROU)z -0.95 / -0.95raw 1.27 / 1.27
Valence (VALN)z -0.38 / -0.21raw 1.91 / 2.09
Recording quality (RCQL)z +1.00 / +1.16raw 3.64 / 3.79
Background noise (BKGN)z +1.56 / +1.33raw 3.89 / 3.72
talking speedz -0.14 / +0.24raw 8.49 / 9.94
Characters per second (CPS): 9.58 raw · z +0.15

P3 metadata teacher estimate; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8340.911
Identity · 250D0.8540.890

Vocal-burst regions

Known
Tiny
Base
0 s12.0 s

Purple marks known inserted burst intervals; other vocal sounds may be unlabeled.

Tiny: 1 regions, count-head estimate 0.92; Base: 1 regions, count-head estimate 1.09.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
Tiny17.848.160.32
Base17.828.160.34

Known inserted intervals: 7.85–8.16 s

All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.551+0.960.983+1.710.0070.570
1Amusementemo_Amusement0.020+0.020.013+0.02-0.0010.885
2Angeremo_Anger0.066+0.100.012+0.010.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.046+0.07-0.005-0.020.0050.572
4Aweemo_Awe0.028+0.070.022+0.050.0000.427
5Bitternessemo_Bitterness0.334+0.770.625+1.470.0130.415
6Concentrationemo_Concentration1.521+1.201.561+1.250.6340.741
7Confusionemo_Confusion1.493+2.761.626+3.010.0010.540
8Contemplationemo_Contemplation1.987+3.192.236+3.600.0820.598
9Contemptemo_Contempt0.032+0.07-0.020-0.04-0.0000.464
10Contentmentemo_Contentment0.279+0.540.396+0.770.0000.514
11Disappointmentemo_Disappointment1.673+2.391.706+2.440.0000.700
12Disgustemo_Disgust0.057+0.100.023+0.030.0070.489
13Distressemo_Distress1.717+2.091.596+1.940.0000.821
14Doubtemo_Doubt1.190+2.541.231+2.630.0060.466
15Elationemo_Elation-0.073-0.12-0.062-0.100.0120.728
16Embarrassmentemo_Embarrassment0.138+0.220.180+0.320.0430.430
17Emotional Numbnessemo_Emotional_Numbness1.303+1.421.412+1.540.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.862+1.550.557+0.980.0280.539
19Fearemo_Fear0.663+1.060.329+0.520.0150.610
20Helplessnessemo_Helplessness1.781+2.561.620+2.33-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism0.077-0.050.092-0.030.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability0.268+0.170.129-0.030.1490.705
23Infatuationemo_Infatuation0.468+0.911.015+1.93-0.0150.533
24Interestemo_Interest1.405-0.201.438-0.171.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness0.008+0.060.038+0.11-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.236+0.400.177+0.270.0590.442
27Longingemo_Longing1.856+2.801.637+2.450.0870.632
28Malevolence Maliceemo_Malevolence_Malice0.096+0.080.027-0.050.0550.534
29Painemo_Pain0.447+0.770.327+0.57-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.013-0.010.016-0.000.0170.614
31Prideemo_Pride0.031+0.060.024+0.040.0020.509
32Reliefemo_Relief0.910+1.290.899+1.270.0330.681
33Sadnessemo_Sadness2.513+3.312.596+3.420.0000.759
34Sexual Lustemo_Sexual_Lust0.155+0.250.189+0.320.0300.494
35Shameemo_Shame1.014+1.961.132+2.190.0130.510
36Sournessemo_Sourness-0.003+0.01-0.004+0.00-0.0050.439
37Teasingemo_Teasing0.018+0.03-0.031-0.060.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.176+0.190.429+0.560.0410.697
39Triumphemo_Triumph0.078+0.160.066+0.14-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg4.227+1.004.512+1.232.9841.238
41Arousal (AROU)vn_AROU_reg1.266-0.951.266-0.952.3591.147
42Arousal shift (ARSH)vn_ARSH_reg1.831-0.861.547-1.192.5800.867
43Vocal attack (ATCK)vn_ATCK_reg1.224-0.721.049-0.862.1621.295
44Background noise (BKGN)vn_BKGN_reg3.887+1.563.716+1.332.7030.761
45Brightness (BRGT)vn_BRGT_reg1.849-0.181.758-0.291.9910.813
46Speech chunking (CHNK)vn_CHNK_reg1.346-0.401.393-0.371.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg3.396+1.463.403+1.472.0180.945
48Cognitive load (COGL)vn_COGL_reg3.224+1.072.997+0.852.0941.061
49Dynamic arc (DARC)vn_DARC_reg1.288-0.241.243-0.271.6151.380
50Disfluency (DFLU)vn_DFLU_reg2.727+0.202.891+0.342.4881.184
51Emphasis (EMPH)vn_EMPH_reg1.992-0.402.164-0.292.6071.524
52Esthetics (ESTH)vn_ESTH_reg3.128+0.933.384+1.262.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.126+0.120.218+0.410.0860.322
54Focus (FOCS)vn_FOCS_reg1.045-1.181.097-1.132.3441.102
55Fullness (FULL)vn_FULL_reg2.255+0.542.136+0.401.7780.889
56Perceived gender (GEND)vn_GEND_reg1.642-1.521.505-1.614.0551.588
57Harmonicity (HARM)vn_HARM_reg1.573-1.091.534-1.132.4800.836
58Metallic character (METL)vn_METL_reg1.220-0.421.097-0.551.6210.958
59Pitch range (RANG)vn_RANG_reg2.646-0.302.693-0.252.9140.897
60Recording quality (RCQL)vn_RCQL_reg3.645+1.003.785+1.162.7540.888
61Vocal register (REGS)vn_REGS_reg3.680+1.963.817+2.061.1981.268
62Respiration (RESP)vn_RESP_reg3.050+0.353.081+0.382.6801.056
63Roughness (ROUG)vn_ROUG_reg1.258-0.871.052-1.112.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg1.610-0.621.477-0.782.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg2.973+1.002.948+0.972.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg2.391+0.482.359+0.451.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg1.852+0.501.800+0.431.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg3.601+0.773.572+0.752.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg2.119-0.092.017-0.222.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg2.713+0.802.697+0.782.0700.803
71Smoothness (SMTH)vn_SMTH_reg2.106-0.032.186+0.042.1431.093
72Stance (STNC)vn_STNC_reg1.693-0.831.409-1.082.6331.137
73Speech structure (STRU)vn_STRU_reg2.765+0.662.861+0.762.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg4.851+2.475.023+2.591.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg1.006-0.080.840-0.241.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg1.819+0.181.900+0.241.5601.401
77Casual style (S_CASU)vn_S_CASU_reg2.205-1.242.534-0.983.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg2.171-0.322.212-0.292.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg1.947-0.671.827-0.782.7211.148
80Formal style (S_FORM)vn_S_FORM_reg2.301+1.082.226+1.011.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg4.177+1.054.415+1.242.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg3.975+1.853.975+1.851.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg1.249+1.481.267+1.510.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg1.202-0.871.207-0.872.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg-0.300-1.17-0.490-1.321.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg2.952+0.343.246+0.562.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg2.061+1.152.089+1.180.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg5.364+2.335.714+2.551.5971.618
89Tempo (TEMP)vn_TEMP_reg1.013-0.931.055-0.892.1171.190
90Tension (TENS)vn_TENS_reg1.420-0.161.312-0.291.5590.852
91Valence (VALN)vn_VALN_reg1.906-0.382.091-0.212.3281.118
92Valence shift (VALS)vn_VALS_reg2.075-0.412.410-0.102.5181.079
93Velocity flux (VFLX)vn_VFLX_reg1.410-0.701.594-0.552.2561.208
94Volatility (VOLT)vn_VOLT_reg1.836-0.851.809-0.882.5860.886
95Vulnerability (VULN)vn_VULN_reg3.344+0.713.862+1.232.6480.984
96Warmth (WARM)vn_WARM_reg2.742+0.182.819+0.272.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_60.955-0.850.930-0.862.3321.621
98Vocal-burst blend (0–10)blend_0_100.083-2.410.168-2.386.6762.738
99Legacy R_quality (no P3 test labels)R_quality0.190-2.680.190-2.681.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.172-0.273.184-0.233.2500.293
101Arousaleiv_extra:Arousal1.028-0.601.072-0.501.2890.435
102Authenticityeiv_extra:Authenticity2.961+1.923.031+2.382.6670.153
103Background Noiseeiv_extra:Background_Noise0.481-1.240.146-2.500.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.589+1.811.596+1.841.0400.303
105Gendereiv_extra:Gender1.582+2.201.488+2.11-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched1.710-0.471.712-0.461.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive1.984+0.272.096+0.501.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.371+0.682.450+0.922.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous0.656-0.800.652-0.820.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh0.915+1.980.905+1.950.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant0.279-0.300.340-0.110.3760.324
112Valenceeiv_extra:Valence-0.552-1.46-1.014-2.680.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.150-1.291.092-1.431.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold0.040-0.040.113+0.210.0510.300
115score background qualityeiv_extra:score_background_quality3.361+0.093.203-0.143.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.739+0.774.719+0.704.5070.303
117score overall qualityeiv_extra:score_overall_quality2.905+0.502.722+0.142.6520.507
118score speech qualityeiv_extra:score_speech_quality1.807-0.011.811+0.021.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.455+0.645.455+0.644.7991.018
120Content usefulness (CU)audiobox:CU5.744+0.105.939+0.295.6491.001
121Production complexity (PC)audiobox:PC2.729+0.252.754+0.262.3311.617
122Production quality (PQ)audiobox:PQ7.036-0.017.210+0.197.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw4.040+0.214.150+0.373.9000.669
124Background quality, rawdnsmos:BAK_raw3.452+0.213.575+0.323.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.275+0.233.459+0.463.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.571+0.173.632+0.293.4880.501
127Background quality, calibrateddnsmos:BAK3.479+0.103.562+0.193.3890.928
128Overall quality, calibrateddnsmos:OVRL2.960+0.203.079+0.392.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.609+0.383.634+0.433.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.652+1.010.737+1.140.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection1.033+0.151.070+0.220.9470.554
132Agevoiceclap_attribute:Age3.208-0.043.226-0.013.2290.547
133Amusementvoiceclap_attribute:Amusement-0.163-0.97-0.123-0.900.3920.572
134Angervoiceclap_attribute:Anger0.605-0.770.591-0.790.9920.506
135Arousalvoiceclap_attribute:Arousal0.633-1.260.754-1.071.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise0.045-0.970.223-0.600.5090.479
137Authenticityvoiceclap_attribute:Authenticity3.083+1.873.155+2.312.7700.167
138Awevoiceclap_attribute:Awe0.476-0.240.484-0.210.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.347-0.840.335-0.880.5670.264
140Bitternessvoiceclap_attribute:Bitterness0.646+0.230.679+0.370.5920.236
141Concentrationvoiceclap_attribute:Concentration1.857+1.211.770+1.051.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.619+1.171.676+1.301.0690.468
143Confusionvoiceclap_attribute:Confusion1.090+1.341.032+1.170.6430.334
144Contemplationvoiceclap_attribute:Contemplation1.988+2.982.019+3.060.7800.406
145Contemptvoiceclap_attribute:Contempt0.062-1.910.040-1.970.7090.339
146Contentmentvoiceclap_attribute:Contentment0.658+0.230.913+1.060.5880.307
147Disappointmentvoiceclap_attribute:Disappointment0.983+0.770.966+0.710.7600.291
148Disgustvoiceclap_attribute:Disgust0.224-1.930.201-2.050.6250.207
149Distressvoiceclap_attribute:Distress1.237+0.891.353+1.220.9270.350
150Doubtvoiceclap_attribute:Doubt0.830+1.570.799+1.400.5350.188
151Elationvoiceclap_attribute:Elation0.106-1.090.330-0.530.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.190-0.370.198-0.350.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness1.046+1.440.922+0.970.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.966+1.930.990+2.020.4620.261
155Fearvoiceclap_attribute:Fear0.656+0.180.647+0.140.6180.212
156Gendervoiceclap_attribute:Gender1.191+1.341.340+1.48-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness1.151+1.841.252+2.210.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched1.843+0.011.775-0.171.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism0.432-0.690.554-0.380.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability0.405-1.200.430-1.151.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.578-0.090.640+0.100.6080.332
162Interestvoiceclap_attribute:Interest1.279-0.721.386-0.481.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness0.129-0.800.240-0.570.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.462-0.250.441-0.330.5230.245
165Longingvoiceclap_attribute:Longing1.027+2.231.037+2.270.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice0.295-1.560.243-1.700.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive1.721-0.091.944+0.281.7750.594
168Painvoiceclap_attribute:Pain0.610+1.050.638+1.180.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.222-0.930.413-0.300.5030.301
170Pridevoiceclap_attribute:Pride0.658-0.350.604-0.510.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality2.589+0.842.603+0.882.3130.328
172Reliefvoiceclap_attribute:Relief1.041+0.041.142+0.281.0250.415
173Sadnessvoiceclap_attribute:Sadness1.501+2.391.569+2.610.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous0.233-1.000.278-0.930.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.039-1.420.147-1.170.6350.419
176Shamevoiceclap_attribute:Shame1.035+1.801.112+2.070.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh0.872+1.980.999+2.330.1600.360
178Sournessvoiceclap_attribute:Sourness0.328-1.300.290-1.430.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant0.242-1.060.211-1.150.6230.359
180Teasingvoiceclap_attribute:Teasing-0.035-1.52-0.053-1.570.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.532+0.281.784+0.761.3840.529
182Triumphvoiceclap_attribute:Triumph0.266-1.250.399-0.880.7160.359
183Valencevoiceclap_attribute:Valence-0.403-0.62-0.273-0.42-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached1.528-0.201.084-1.001.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold0.204+0.670.328+0.99-0.0660.400
186durationvoiceclap_attribute:duration12.965+0.3213.330+0.3810.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.454+0.423.474+0.463.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.673+0.124.652-0.034.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.879+0.592.907+0.662.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.870-0.301.874-0.241.8910.069
191talking speedvoiceclap_attribute:talking_speed8.492-0.149.943+0.249.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.02630.09660.1611
10.19990.22100.1793
20.04400.05440.0296
3-0.0319-0.0116-0.0215
4-0.0853-0.0855-0.0750
5-0.0511-0.0364-0.0582
60.04860.07080.0993
70.10990.08060.0546
80.13220.10480.1009
90.00730.0146-0.0192
10-0.0965-0.0778-0.0742
11-0.0189-0.0520-0.0802
12-0.1092-0.0860-0.0803
130.08730.07530.1320
140.08960.0463-0.0317
150.03500.07330.0736
16-0.0830-0.0753-0.0668
17-0.0830-0.1138-0.0830
18-0.0915-0.0782-0.0736
19-0.0479-0.0675-0.0755
20-0.0162-0.0467-0.0766
210.17380.19320.1622
220.07800.11870.0976
230.05570.00930.0466
240.04790.01900.0194
250.01680.0270-0.0120
26-0.1092-0.0962-0.0748
270.02690.01840.1087
28-0.1030-0.1064-0.0880
29-0.0177-0.0610-0.0854
300.01920.06430.1018
31-0.1115-0.0864-0.0483
320.02390.0307-0.0126
330.17760.15060.1261
34-0.0738-0.0802-0.0715
350.06730.02110.0641
36-0.0600-0.1011-0.0858
370.13300.06340.0742
380.06730.10720.1492
390.06070.03090.0169
40-0.02240.00160.0150
410.03250.04320.0723
42-0.1084-0.0855-0.0805
43-0.0903-0.0872-0.0821
440.01280.0393-0.0023
45-0.0923-0.0233-0.0306
46-0.0421-0.0790-0.0753
47-0.0999-0.1048-0.0807
48-0.1099-0.0790-0.0779
49-0.01950.0017-0.0094
500.09460.07980.0471
51-0.0198-0.0008-0.0416
520.06800.0245-0.0476
53-0.0279-0.0465-0.0153
54-0.0953-0.0843-0.0893
550.01920.0262-0.0252
56-0.0834-0.0839-0.0757
570.13760.19160.1167
58-0.0684-0.0778-0.0141
590.05540.07860.1284
600.06840.13830.0925
610.11990.06840.0249
62-0.0092-0.00890.0300
630.04610.11300.1554
640.11690.13510.1783
65-0.0934-0.0745-0.0651
660.01990.06100.0955
67-0.0880-0.0688-0.0770
680.16760.14330.1152
690.12530.06790.0305
700.17610.13340.1355
710.05500.05690.0422
720.19370.17850.1565
730.07500.07040.1494
740.0016-0.0197-0.0657
750.23530.18420.2090
760.13690.12280.0224
77-0.0846-0.0860-0.0729
78-0.0765-0.0720-0.0710
790.10530.0235-0.0140
80-0.1061-0.1089-0.0723
810.12380.12360.0869
82-0.0115-0.0301-0.0657
83-0.0673-0.0446-0.0817
840.05000.05890.1469
85-0.0361-0.0194-0.0238
860.13760.17030.1724
870.01250.00150.0283
88-0.1030-0.0741-0.0742
89-0.1046-0.0802-0.0732
900.05840.11790.1460
91-0.0081-0.00400.0518
92-0.0263-0.0475-0.0775
930.03400.05010.0510
940.13150.12520.1520
95-0.0329-0.0350-0.0487
96-0.0607-0.0630-0.0275
970.08420.08190.1125
98-0.0119-0.0610-0.0806
990.22450.21610.1515
100-0.0088-0.02260.0395
101-0.0792-0.0851-0.0771
1020.04170.01820.0865
103-0.0642-0.0872-0.0842
1040.04610.04440.0493
105-0.0619-0.1154-0.0770
106-0.1007-0.1044-0.0959
107-0.0780-0.0925-0.0743
108-0.1046-0.0933-0.0865
1090.10460.14160.1666
110-0.0429-0.0338-0.0928
1110.06840.06880.0556
112-0.0323-0.0422-0.0348
113-0.0604-0.0356-0.0635
1140.03710.0178-0.0229
1150.13690.10150.0621
1160.09340.10810.0594
117-0.0727-0.0868-0.0716
118-0.0926-0.0815-0.0777
1190.00510.0323-0.0239
1200.16990.19970.1801
1210.0054-0.0332-0.0402
122-0.0911-0.1072-0.0820
123-0.0515-0.0259-0.0066
124-0.03270.00870.1066
125-0.01750.02600.1129
1260.03860.03170.0264
1270.06800.13750.0879
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.0312-0.0427-0.0451
1-0.0425-0.0383-0.0400
20.01350.05440.0551
3-0.0359-0.0445-0.0541
40.04570.05830.0464
50.07670.05930.1240
60.08140.09320.0453
7-0.0718-0.0482-0.0419
80.02740.05750.0558
90.04920.08800.0726
10-0.0154-0.0019-0.0274
11-0.0533-0.0417-0.0560
120.11420.10720.0855
130.18040.22060.2203
14-0.0140-0.0070-0.0065
150.10720.09890.0714
160.0082-0.0136-0.0238
17-0.02970.0065-0.0497
18-0.0046-0.0002-0.0134
19-0.03900.00510.0157
200.02870.03470.0868
21-0.0277-0.0208-0.0384
22-0.0413-0.0513-0.0429
23-0.0630-0.0435-0.0400
240.09720.13100.1317
25-0.04160.01290.0322
260.11360.11760.2103
270.02500.00950.0442
280.04920.05390.0233
29-0.0387-0.0585-0.0398
30-0.0278-0.0328-0.0360
310.02170.0179-0.0498
320.14590.17610.1725
330.05480.02600.0210
340.08790.07040.0215
35-0.0274-0.0300-0.0422
360.11130.06160.0309
37-0.0280-0.0267-0.0286
38-0.0419-0.0373-0.0574
390.03270.0211-0.0038
400.01960.02210.0231
41-0.0338-0.0030-0.0475
420.14410.15430.0992
43-0.0803-0.1031-0.0489
44-0.0038-0.0242-0.0463
450.00940.0092-0.0241
46-0.0030-0.03240.0212
470.12770.11240.0736
480.01710.00650.0288
490.08670.08340.0956
500.11070.09680.1833
510.05620.05850.0218
52-0.0723-0.0520-0.0408
530.05540.07720.0415
54-0.0838-0.0570-0.0699
550.00970.02240.0093
560.02960.0234-0.0037
570.0082-0.00420.0408
58-0.0372-0.0443-0.0397
590.07670.07350.0804
60-0.02200.0016-0.0403
61-0.0398-0.0329-0.0384
620.06770.09890.0620
63-0.0392-0.0414-0.0290
64-0.0008-0.0445-0.0461
65-0.0199-0.0199-0.0226
660.00040.04140.0138
670.05360.0236-0.0316
680.06740.05980.0939
69-0.02180.00410.0272
70-0.0091-0.0183-0.0360
71-0.0350-0.0362-0.0271
720.01300.02280.0177
730.06650.05460.0160
74-0.0521-0.0629-0.0446
75-0.0155-0.0448-0.0602
760.05210.04950.0711
77-0.1318-0.1321-0.0878
780.10430.09060.1562
79-0.0174-0.0134-0.0073
80-0.0791-0.0533-0.0439
81-0.0384-0.0241-0.0495
82-0.0011-0.0126-0.0397
830.09610.09790.1351
84-0.02830.0131-0.0623
85-0.0469-0.0225-0.0327
86-0.0097-0.0198-0.0281
87-0.0193-0.0350-0.0352
880.00950.0080-0.0205
89-0.0492-0.0386-0.0399
900.10720.02860.0510
910.08080.08540.1904
92-0.0246-0.0443-0.0382
93-0.0685-0.0777-0.0501
94-0.0310-0.02190.0462
950.04100.06890.0422
960.06590.06580.1094
97-0.0750-0.0510-0.0520
98-0.00650.03780.0027
99-0.1125-0.0999-0.0553
1000.20150.15640.1852
1010.10660.14910.0399
1020.04070.0344-0.0033
1030.03540.0193-0.0577
1040.0212-0.0035-0.0223
105-0.0028-0.00660.0403
106-0.0556-0.0699-0.0340
1070.03870.05200.0185
108-0.0312-0.0396-0.0329
109-0.0849-0.0668-0.0425
110-0.0177-0.0277-0.0374
1110.09140.10100.0471
1120.04190.08130.0581
1130.08900.03040.0163
114-0.0371-0.0557-0.0616
115-0.0808-0.0849-0.0313
1160.07730.09480.0648
117-0.0735-0.0803-0.0721
1180.01420.01110.0649
1190.05970.05900.1024
120-0.1060-0.0963-0.0563
1210.07380.09170.0620
122-0.0501-0.0448-0.0417
123-0.0316-0.0236-0.0452
1240.15820.14600.0967
125-0.01210.0220-0.0127
1260.0436-0.01620.0223
127-0.0139-0.0407-0.0454
128-0.0335-0.0435-0.0550
1290.0049-0.0197-0.0420
1300.03810.07250.1056
131-0.0510-0.0564-0.0386
132-0.0315-0.0303-0.0468
133-0.0469-0.0339-0.0338
1340.13360.08700.0740
1350.00650.02410.0095
1360.09200.07510.0381
137-0.0199-0.0057-0.0108
138-0.0463-0.0337-0.0389
139-0.0250-0.0181-0.0204
1400.08670.12220.1057
141-0.0682-0.0564-0.0437
1420.03080.04170.0629
1430.08790.09680.1080
144-0.03750.01700.0214
145-0.0321-0.0180-0.0271
146-0.0195-0.0177-0.0369
1470.0110-0.0438-0.0392
1480.0027-0.0047-0.0318
1490.17220.13780.1357
150-0.0738-0.0614-0.0435
151-0.0059-0.0229-0.0465
152-0.0398-0.0684-0.0490
153-0.0124-0.0430-0.0534
1540.02370.03810.0160
1550.07730.07980.0476
1560.02080.03110.0529
1570.02460.01500.0247
1580.01510.0049-0.0108
159-0.0726-0.0518-0.0517
1600.02740.02540.0525
161-0.0265-0.00770.0016
162-0.0183-0.0153-0.0157
1630.03900.04890.0652
1640.00890.01470.0308
165-0.1002-0.0829-0.0632
1660.01550.05050.0494
1670.04250.00750.0199
168-0.0972-0.0772-0.0624
1690.10660.07770.0930
1700.00150.00230.0480
1710.06650.08230.0379
172-0.0049-0.0042-0.0179
173-0.0302-0.0143-0.0518
174-0.0527-0.0370-0.0453
175-0.1172-0.1181-0.0697
176-0.0321-0.0502-0.0375
177-0.0181-0.0458-0.0364
1780.03210.02510.0010
179-0.00260.02890.0128
1800.09780.11080.1208
1810.01360.02540.0442
182-0.0483-0.0419-0.0381
183-0.0580-0.0709-0.0832
184-0.0554-0.0585-0.0449
185-0.0271-0.0306-0.0148
1860.06000.04480.0566
187-0.0603-0.02370.0195
188-0.0160-0.0227-0.0132
1890.05650.08700.0551
1900.05360.04760.0525
191-0.0171-0.0787-0.0590
1920.06880.0357-0.0016
193-0.0656-0.0792-0.0465
194-0.0090-0.0120-0.0174
195-0.0838-0.0645-0.0420
1960.05620.07250.1050
197-0.0332-0.02010.0068
1980.02890.00620.0131
199-0.0065-0.0061-0.0482
200-0.0738-0.0678-0.0511
2010.0100-0.0232-0.0416
202-0.0524-0.0142-0.0325
203-0.0130-0.0260-0.0192
2040.02470.01540.0261
2050.01440.03470.0200
2060.07560.06190.0247
207-0.0647-0.0603-0.0597
208-0.0289-0.0079-0.0262
209-0.0161-0.0265-0.0482
2100.10840.10560.1292
211-0.0372-0.0234-0.0339
2120.03920.04350.0568
2130.12010.07980.1010
214-0.0283-0.0451-0.0490
215-0.0398-0.0337-0.0371
2160.24490.28170.2708
2170.0025-0.0320-0.0495
218-0.0738-0.0813-0.0512
219-0.0472-0.0197-0.0334
220-0.0779-0.1046-0.0487
2210.09260.12330.1186
222-0.0231-0.0280-0.0100
223-0.0253-0.0241-0.0354
224-0.0262-0.0272-0.0403
2250.11720.12380.1154
2260.10720.07250.0562
227-0.0398-0.0482-0.0583
2280.09670.08440.1165
229-0.0271-0.0401-0.0275
230-0.0820-0.0629-0.0450
2310.06060.05800.0635
2320.00360.0020-0.0528
233-0.0486-0.0564-0.0001
234-0.00480.0211-0.0184
235-0.0375-0.0417-0.0363
236-0.0283-0.0339-0.0379
237-0.0195-0.01070.0158
238-0.00120.00760.0362
239-0.0492-0.0448-0.0456
2400.0058-0.00360.0043
2410.07380.03370.0125
242-0.0568-0.0388-0.0211
243-0.0603-0.0572-0.0567
244-0.0195-0.00420.0313
245-0.0116-0.0059-0.0420
246-0.0545-0.0523-0.0357
2470.0004-0.0165-0.0372
248-0.0609-0.0720-0.0505
2490.10130.09790.0723
Source: P3 source-disjoint test split · p3_02004911 · SHA-256 63d4f9c77d41cc0c…
p06Speech + burstsP3 curated burst test

Triumph + 1 known burst(s)

Synthetic speech with inserted bursts · P3 synthetic speech/burst mixture · 16.0 s

Tiny Base · values are model estimates, not ground truth

Strongest emotions relative to training

Anger
z +2.31 / +2.11raw 1.42 / 1.29
Impatience and Irritability
z +2.11 / +1.89raw 1.64 / 1.48
Triumph
z +2.09 / +1.12raw 1.08 / 0.57
Malevolence Malice
z +1.93 / +1.96raw 1.09 / 1.10
Teasing
z +1.71 / +1.91raw 0.93 / 1.04
Hope Enthusiasm Optimism
z +1.61 / +0.91raw 1.34 / 0.81
Disgust
z +1.04 / +1.56raw 0.52 / 0.77

The center of each bar is the S9 teacher median. Positive z is to the right; negative z is to the left. Ranking uses z, not raw teacher units.

Strongest VoiceNet dimensions relative to training

Tension (TENS)
z +2.91 / +2.70raw 4.04 / 3.86
Cartoonish style (S_CART)
z +2.82 / +2.79raw 5.51 / 5.47
Dynamic arc (DARC)
z +2.37 / +1.96raw 4.89 / 4.33
Pitch range (RANG)
z +2.26 / +2.06raw 4.95 / 4.77
Arousal (AROU)
z +2.22 / +1.92raw 4.90 / 4.56
Stance (STNC)
z +2.19 / +2.03raw 5.12 / 4.94
Roughness (ROUG)
z +2.17 / +2.02raw 3.96 / 3.82

All 57 VoiceNet predictions, including raw units, are in the full table below. These seven are ranked by z; the top values may differ from the strongest raw dimensions.

Voice, speed and quality scores in context

Genuineness (0–6)z +0.74 / +0.56raw 3.53 / 3.24
Vocal-burst blend (0–10)z -0.64 / -0.74raw 4.93 / 4.66
Content enjoyment (CE)z +0.58 / +0.33raw 5.39 / 5.14
Production quality (PQ)z +0.68 / +0.75raw 7.64 / 7.70
Speech signal quality, calibratedz +0.28 / +0.36raw 3.63 / 3.67
Background quality, calibratedz +0.34 / +0.74raw 3.70 / 4.08
Overall quality, calibratedz +0.45 / +0.84raw 3.11 / 3.35
P.808 MOS estimatez -0.08 / +0.25raw 3.38 / 3.54
score content enjoymentz +0.86 / +0.84raw 4.77 / 4.76
score overall qualityz +0.39 / +0.50raw 2.85 / 2.91
Background Noisez +0.11 / +0.25raw 0.84 / 0.87
Tempo (TEMP)z +0.87 / +0.56raw 3.15 / 2.78
Arousal (AROU)z +2.22 / +1.92raw 4.90 / 4.56
Valence (VALN)z -0.22 / -0.38raw 2.08 / 1.90
Recording quality (RCQL)z -1.67 / -1.71raw 1.27 / 1.23
Background noise (BKGN)z -0.67 / -0.69raw 2.20 / 2.18
talking speedz -1.21 / -1.33raw 4.50 / 4.03
Characters per second (CPS): 7.26 raw · z -0.47

P3 metadata teacher estimate; not a Whisper output. Reference: 86,334 P3 training teacher values, median 9.03, SD 3.76. Cross-language character counts depend on writing system; compare with care.

Orange speaker embedding reconstruction

Cosine against Orange on this same clip (higher means closer to that teacher vector):

EmbeddingTinyBase
Timbre · 128D0.8180.884
Identity · 250D0.7810.755

Vocal-burst regions

Known
Tiny
Base
0 s16.0 s

Purple marks known inserted burst intervals; other vocal sounds may be unlabeled.

Tiny: 2 regions, count-head estimate 0.46; Base: 1 regions, count-head estimate 0.80.

Exact predicted start and end times
ModelRegionStart (s)End (s)Duration (s)
Tiny12.002.220.22
Tiny22.243.080.84
Base11.942.941.00

Known inserted intervals: 1.94–2.86 s

All 192 score slots · raw and z values

z = (prediction − training teacher median) / training teacher SD. Higher values are not universally better. R_quality has Ladder training labels but no P3 test labels, so no held-out P3 metric is available for it.

#TargetTiny rawTiny zBase rawBase zTrain medianTrain SD
Emotion scores Forty clip-level Empathic Insight emotion teacher outputs. Each row evaluates one named emotion score.
0Affectionemo_Affection0.804+1.400.478+0.830.0070.570
1Amusementemo_Amusement1.040+1.181.142+1.29-0.0010.885
2Angeremo_Anger1.417+2.311.294+2.110.0040.613
3Astonishment Surpriseemo_Astonishment_Surprise0.272+0.470.331+0.570.0050.572
4Aweemo_Awe0.018+0.040.005+0.010.0000.427
5Bitternessemo_Bitterness0.622+1.460.568+1.330.0130.415
6Concentrationemo_Concentration1.012+0.511.009+0.510.6340.741
7Confusionemo_Confusion0.441+0.820.194+0.360.0010.540
8Contemplationemo_Contemplation0.117+0.060.069-0.020.0820.598
9Contemptemo_Contempt0.602+1.300.641+1.38-0.0000.464
10Contentmentemo_Contentment0.716+1.390.455+0.890.0000.514
11Disappointmentemo_Disappointment0.366+0.520.327+0.470.0000.700
12Disgustemo_Disgust0.515+1.040.770+1.560.0070.489
13Distressemo_Distress0.202+0.250.202+0.250.0000.821
14Doubtemo_Doubt-0.051-0.120.023+0.040.0060.466
15Elationemo_Elation0.996+1.350.622+0.840.0120.728
16Embarrassmentemo_Embarrassment0.212+0.390.425+0.890.0430.430
17Emotional Numbnessemo_Emotional_Numbness-0.045-0.05-0.049-0.060.0040.917
18Fatigue Exhaustionemo_Fatigue_Exhaustion0.018-0.020.049+0.040.0280.539
19Fearemo_Fear-0.032-0.08-0.020-0.060.0150.610
20Helplessnessemo_Helplessness-0.072-0.10-0.031-0.04-0.0040.696
21Hope Enthusiasm Optimismemo_Hope_Enthusiasm_Optimism1.339+1.610.812+0.910.1170.760
22Impatience and Irritabilityemo_Impatience_and_Irritability1.637+2.111.482+1.890.1490.705
23Infatuationemo_Infatuation-0.043-0.050.003+0.03-0.0150.533
24Interestemo_Interest2.195+0.672.110+0.571.5890.907
25Intoxication Altered States of Consciousnessemo_Intoxication_Altered_States_of_Consciousness0.564+1.050.598+1.12-0.0250.559
26Jealousy and Envyemo_Jealousy_and_Envy0.107+0.110.045-0.030.0590.442
27Longingemo_Longing0.058-0.050.050-0.060.0870.632
28Malevolence Maliceemo_Malevolence_Malice1.086+1.931.102+1.960.0550.534
29Painemo_Pain-0.026-0.040.035+0.06-0.0020.579
30Pleasure Ecstasyemo_Pleasure_Ecstasy0.736+1.170.512+0.810.0170.614
31Prideemo_Pride0.744+1.460.372+0.730.0020.509
32Reliefemo_Relief0.108+0.110.067+0.050.0330.681
33Sadnessemo_Sadness0.020+0.030.023+0.030.0000.759
34Sexual Lustemo_Sexual_Lust0.079+0.100.090+0.120.0300.494
35Shameemo_Shame0.120+0.210.144+0.260.0130.510
36Sournessemo_Sourness0.364+0.840.339+0.78-0.0050.439
37Teasingemo_Teasing0.930+1.711.043+1.910.0000.545
38Thankfulness Gratitudeemo_Thankfulness_Gratitude0.713+0.960.469+0.610.0410.697
39Triumphemo_Triumph1.080+2.090.575+1.12-0.0070.521
VoiceNet dimensions Fifty-seven VoiceNet voice, delivery, style, and recording dimensions. Codes are expanded in the target column.
40Perceived voice age (AGEV)vn_AGEV_reg3.122+0.113.380+0.322.9841.238
41Arousal (AROU)vn_AROU_reg4.901+2.224.558+1.922.3591.147
42Arousal shift (ARSH)vn_ARSH_reg3.392+0.943.091+0.592.5800.867
43Vocal attack (ATCK)vn_ATCK_reg4.351+1.694.207+1.582.1621.295
44Background noise (BKGN)vn_BKGN_reg2.197-0.672.179-0.692.7030.761
45Brightness (BRGT)vn_BRGT_reg1.725-0.331.495-0.611.9910.813
46Speech chunking (CHNK)vn_CHNK_reg0.713-0.920.528-1.061.8471.239
47Articulation clarity (CLRT)vn_CLRT_reg2.064+0.052.037+0.022.0180.945
48Cognitive load (COGL)vn_COGL_reg0.786-1.230.753-1.262.0941.061
49Dynamic arc (DARC)vn_DARC_reg4.890+2.374.326+1.961.6151.380
50Disfluency (DFLU)vn_DFLU_reg2.750+0.222.859+0.312.4881.184
51Emphasis (EMPH)vn_EMPH_reg5.271+1.754.854+1.472.6071.524
52Esthetics (ESTH)vn_ESTH_reg0.810-2.040.865-1.972.4020.782
53Content appropriateness (EXPL)vn_EXPL_reg0.385+0.930.345+0.800.0860.322
54Focus (FOCS)vn_FOCS_reg3.941+1.453.342+0.912.3441.102
55Fullness (FULL)vn_FULL_reg1.542-0.271.078-0.791.7780.889
56Perceived gender (GEND)vn_GEND_reg5.088+0.655.213+0.734.0551.588
57Harmonicity (HARM)vn_HARM_reg1.495-1.181.495-1.182.4800.836
58Metallic character (METL)vn_METL_reg3.001+1.442.806+1.241.6210.958
59Pitch range (RANG)vn_RANG_reg4.946+2.264.765+2.062.9140.897
60Recording quality (RCQL)vn_RCQL_reg1.275-1.671.232-1.712.7540.888
61Vocal register (REGS)vn_REGS_reg0.510-0.540.274-0.731.1981.268
62Respiration (RESP)vn_RESP_reg3.185+0.482.966+0.272.6801.056
63Roughness (ROUG)vn_ROUG_reg3.958+2.173.821+2.022.0330.887
64Chest resonance (R_CHST)vn_R_CHST_reg3.262+1.343.016+1.042.1350.844
65Head resonance (R_HEAD)vn_R_HEAD_reg1.749-0.491.535-0.752.1480.822
66Mask resonance (R_MASK)vn_R_MASK_reg2.538+0.632.223+0.311.9081.005
67Mixed resonance (R_MIXD)vn_R_MIXD_reg0.991-0.530.938-0.591.4360.841
68Nasal resonance (R_NASL)vn_R_NASL_reg4.417+1.554.351+1.492.7911.047
69Oral resonance (R_ORAL)vn_R_ORAL_reg2.479+0.402.220+0.052.1840.741
70Throat resonance (R_THRT)vn_R_THRT_reg3.295+1.533.068+1.242.0700.803
71Smoothness (SMTH)vn_SMTH_reg1.627-0.471.442-0.642.1431.093
72Stance (STNC)vn_STNC_reg5.118+2.194.940+2.032.6331.137
73Speech structure (STRU)vn_STRU_reg1.488-0.681.359-0.812.1370.958
74ASMR style (S_ASMR)vn_S_ASMR_reg-0.309-1.19-0.190-1.111.3701.411
75Authoritative style (S_AUTH)vn_S_AUTH_reg2.599+1.452.681+1.531.0881.043
76Cartoonish style (S_CART)vn_S_CART_reg5.509+2.825.466+2.791.5601.401
77Casual style (S_CASU)vn_S_CASU_reg5.953+1.695.475+1.323.7931.278
78Conversational style (S_CONV)vn_S_CONV_reg1.904-0.521.860-0.562.5941.318
79Dramatic style (S_DRAM)vn_S_DRAM_reg5.174+2.144.951+1.942.7211.148
80Formal style (S_FORM)vn_S_FORM_reg-0.314-1.48-0.228-1.391.1951.021
81Monologue style (S_MONO)vn_S_MONO_reg1.344-1.151.528-1.012.8241.284
82Narrator style (S_NARR)vn_S_NARR_reg0.270-0.720.783-0.361.3071.446
83Newsreader style (S_NEWS)vn_S_NEWS_reg0.106-0.530.169-0.420.4080.568
84Playful style (S_PLAY)vn_S_PLAY_reg3.305+0.723.690+1.022.3521.315
85Ranting or angry style (S_RANT)vn_S_RANT_reg3.597+1.853.837+2.041.2131.289
86Storytelling style (S_STRY)vn_S_STRY_reg5.129+1.945.282+2.052.4801.366
87Teacher or didactic style (S_TECH)vn_S_TECH_reg0.153-0.400.729+0.070.6461.227
88Whispered talking style (S_WHIS)vn_S_WHIS_reg1.117-0.301.086-0.321.5971.618
89Tempo (TEMP)vn_TEMP_reg3.147+0.872.781+0.562.1171.190
90Tension (TENS)vn_TENS_reg4.039+2.913.857+2.701.5590.852
91Valence (VALN)vn_VALN_reg2.081-0.221.905-0.382.3281.118
92Valence shift (VALS)vn_VALS_reg1.837-0.631.717-0.742.5181.079
93Velocity flux (VFLX)vn_VFLX_reg1.264-0.820.802-1.202.2561.208
94Volatility (VOLT)vn_VOLT_reg4.394+2.044.238+1.862.5860.886
95Vulnerability (VULN)vn_VULN_reg0.217-2.470.326-2.362.6480.984
96Warmth (WARM)vn_WARM_reg2.068-0.562.004-0.632.5780.909
Genuineness, blend, and legacy quality Genuineness and vocal-burst blend are predicted. R_quality has Ladder training labels but no valid P3 test labels.
97Genuineness (0–6)genuineness_0_63.531+0.743.236+0.562.3321.621
98Vocal-burst blend (0–10)blend_0_104.935-0.644.659-0.746.6762.738
99Legacy R_quality (no P3 test labels)R_quality1.084-0.031.010-0.251.0950.338
Empathic Insight Plus extra scores Nineteen additional clip-level teacher scores, including background noise and content enjoyment.
100Ageeiv_extra:Age3.155-0.333.198-0.183.2500.293
101Arousaleiv_extra:Arousal2.526+2.842.363+2.471.2890.435
102Authenticityeiv_extra:Authenticity2.821+1.002.759+0.602.6670.153
103Background Noiseeiv_extra:Background_Noise0.838+0.110.874+0.250.8080.264
104Confident vs. Hesitanteiv_extra:Confident_vs._Hesitant1.061+0.070.984-0.181.0400.303
105Gendereiv_extra:Gender-1.301-0.68-1.130-0.51-0.6210.999
106High-Pitched vs. Low-Pitchedeiv_extra:High-Pitched_vs._Low-Pitched2.265+1.752.120+1.171.8270.250
107Monotone vs. Expressiveeiv_extra:Monotone_vs._Expressive2.690+1.772.748+1.891.8590.470
108Recording Qualityeiv_extra:Recording_Quality2.070-0.222.052-0.282.1440.332
109Serious vs. Humorouseiv_extra:Serious_vs._Humorous1.360+1.271.440+1.500.9290.340
110Soft vs. Harsheiv_extra:Soft_vs._Harsh-0.128-1.28-0.146-1.330.2810.320
111Submissive vs. Dominanteiv_extra:Submissive_vs._Dominant1.207+2.571.252+2.710.3760.324
112Valenceeiv_extra:Valence0.396+1.030.106+0.270.0050.380
113Vulnerable vs. Emotionally Detachedeiv_extra:Vulnerable_vs._Emotionally_Detached1.299-0.951.258-1.041.7150.437
114Warm vs. Coldeiv_extra:Warm_vs._Cold-0.119-0.57-0.178-0.760.0510.300
115score background qualityeiv_extra:score_background_quality3.472+0.253.594+0.423.3010.693
116score content enjoymenteiv_extra:score_content_enjoyment4.769+0.864.763+0.844.5070.303
117score overall qualityeiv_extra:score_overall_quality2.850+0.392.905+0.502.6520.507
118score speech qualityeiv_extra:score_speech_quality1.812+0.031.847+0.271.8080.146
Meta AudioBox Aesthetics Four AudioBox aesthetic teacher outputs. Expansions follow the official Meta model card.
119Content enjoyment (CE)audiobox:CE5.387+0.585.135+0.334.7991.018
120Content usefulness (CU)audiobox:CU6.253+0.606.210+0.565.6491.001
121Production complexity (PC)audiobox:PC2.028-0.191.934-0.252.3311.617
122Production quality (PQ)audiobox:PQ7.641+0.687.699+0.757.0440.879
Microsoft DNSMOS Seven DNSMOS teacher outputs: raw and calibrated SIG/BAK/OVRL estimates plus P.808 MOS.
123Speech signal quality, rawdnsmos:SIG_raw4.121+0.334.210+0.463.9000.669
124Background quality, rawdnsmos:BAK_raw3.590+0.344.250+0.923.2111.124
125Overall quality, rawdnsmos:OVRL_raw3.465+0.473.859+0.953.0850.812
126Speech signal quality, calibrateddnsmos:SIG3.628+0.283.669+0.363.4880.501
127Background quality, calibrateddnsmos:BAK3.702+0.344.075+0.743.3890.928
128Overall quality, calibrateddnsmos:OVRL3.115+0.453.349+0.842.8400.606
129P.808 MOS estimatednsmos:P808_MOS3.379-0.083.545+0.253.4210.496
Vocal-burst count The regression target is log(1 + count). A separate table also reports count error after inverse transformation.
130Vocal-burst count, log(1 + count)burst_count_log1p0.380+0.590.590+0.910.0000.647
VoiceCLAP attributes Sixty-one individual VoiceCLAP attribute teacher scores. The raw 768D VoiceCLAP embedding is stored in sidecars but is not a model output.
131Affectionvoiceclap_attribute:Affection0.758-0.340.841-0.190.9470.554
132Agevoiceclap_attribute:Age2.625-1.113.025-0.373.2290.547
133Amusementvoiceclap_attribute:Amusement1.303+1.591.321+1.620.3920.572
134Angervoiceclap_attribute:Anger1.929+1.851.810+1.620.9920.506
135Arousalvoiceclap_attribute:Arousal2.894+2.372.695+2.051.4200.622
136Astonishment Surprisevoiceclap_attribute:Astonishment_Surprise1.313+1.681.049+1.130.5090.479
137Authenticityvoiceclap_attribute:Authenticity2.797+0.162.807+0.222.7700.167
138Awevoiceclap_attribute:Awe0.546+0.030.423-0.450.5390.259
139Background Noisevoiceclap_attribute:Background_Noise0.905+1.280.841+1.040.5670.264
140Bitternessvoiceclap_attribute:Bitterness1.075+2.051.065+2.010.5920.236
141Concentrationvoiceclap_attribute:Concentration0.601-1.160.634-1.091.2140.531
142Confident vs. Hesitantvoiceclap_attribute:Confident_vs._Hesitant1.336+0.571.206+0.291.0690.468
143Confusionvoiceclap_attribute:Confusion0.816+0.520.844+0.600.6430.334
144Contemplationvoiceclap_attribute:Contemplation0.303-1.180.338-1.090.7800.406
145Contemptvoiceclap_attribute:Contempt1.250+1.601.298+1.740.7090.339
146Contentmentvoiceclap_attribute:Contentment0.468-0.390.571-0.050.5880.307
147Disappointmentvoiceclap_attribute:Disappointment1.251+1.691.048+0.990.7600.291
148Disgustvoiceclap_attribute:Disgust0.908+1.370.829+0.990.6250.207
149Distressvoiceclap_attribute:Distress1.244+0.911.013+0.250.9270.350
150Doubtvoiceclap_attribute:Doubt0.408-0.670.425-0.580.5350.188
151Elationvoiceclap_attribute:Elation1.255+1.801.218+1.710.5390.397
152Embarrassmentvoiceclap_attribute:Embarrassment0.447+0.450.576+0.860.3070.314
153Emotional Numbnessvoiceclap_attribute:Emotional_Numbness0.129-2.010.092-2.150.6640.266
154Fatigue Exhaustionvoiceclap_attribute:Fatigue_Exhaustion0.162-1.150.249-0.820.4620.261
155Fearvoiceclap_attribute:Fear0.777+0.750.575-0.200.6180.212
156Gendervoiceclap_attribute:Gender-1.462-1.30-1.140-0.98-0.1511.005
157Helplessnessvoiceclap_attribute:Helplessness0.568-0.330.476-0.680.6580.268
158High-Pitched vs. Low-Pitchedvoiceclap_attribute:High-Pitched_vs._Low-Pitched2.321+1.282.251+1.101.8400.375
159Hope Enthusiasm Optimismvoiceclap_attribute:Hope_Enthusiasm_Optimism1.398+1.751.280+1.450.7050.396
160Impatience and Irritabilityvoiceclap_attribute:Impatience_and_Irritability2.382+2.461.971+1.701.0520.541
161Infatuationvoiceclap_attribute:Infatuation0.362-0.740.498-0.330.6080.332
162Interestvoiceclap_attribute:Interest1.909+0.711.859+0.591.5980.440
163Intoxication Altered States of Consciousnessvoiceclap_attribute:Intoxication_Altered_States_of_Consciousness1.112+1.321.068+1.220.5020.463
164Jealousy & Envyvoiceclap_attribute:Jealousy_&_Envy0.666+0.590.728+0.840.5230.245
165Longingvoiceclap_attribute:Longing0.287-1.140.269-1.220.5380.220
166Malevolence Malicevoiceclap_attribute:Malevolence_Malice1.517+1.981.365+1.540.8330.346
167Monotone vs. Expressivevoiceclap_attribute:Monotone_vs._Expressive2.898+1.892.921+1.931.7750.594
168Painvoiceclap_attribute:Pain0.480+0.450.404+0.100.3810.218
169Pleasure Ecstasyvoiceclap_attribute:Pleasure_Ecstasy0.772+0.890.912+1.360.5030.301
170Pridevoiceclap_attribute:Pride1.226+1.281.324+1.560.7800.349
171Recording Qualityvoiceclap_attribute:Recording_Quality1.772-1.651.956-1.092.3130.328
172Reliefvoiceclap_attribute:Relief0.647-0.910.559-1.121.0250.415
173Sadnessvoiceclap_attribute:Sadness0.604-0.470.567-0.590.7520.313
174Serious vs. Humorousvoiceclap_attribute:Serious_vs._Humorous1.378+1.011.267+0.820.8030.568
175Sexual Lustvoiceclap_attribute:Sexual_Lust0.561-0.180.599-0.090.6350.419
176Shamevoiceclap_attribute:Shame0.675+0.520.709+0.650.5260.283
177Soft vs. Harshvoiceclap_attribute:Soft_vs._Harsh-0.361-1.44-0.299-1.270.1600.360
178Sournessvoiceclap_attribute:Sourness0.931+0.781.084+1.300.7060.291
179Submissive vs. Dominantvoiceclap_attribute:Submissive_vs._Dominant1.220+1.671.220+1.670.6230.359
180Teasingvoiceclap_attribute:Teasing1.337+2.101.456+2.420.5400.379
181Thankfulness Gratitudevoiceclap_attribute:Thankfulness_Gratitude1.408+0.051.640+0.481.3840.529
182Triumphvoiceclap_attribute:Triumph1.611+2.491.544+2.310.7160.359
183Valencevoiceclap_attribute:Valence-0.874-1.37-0.618-0.96-0.0070.635
184Vulnerable vs. Emotionally Detachedvoiceclap_attribute:Vulnerable_vs._Emotionally_Detached1.066-1.031.040-1.081.6380.556
185Warm vs. Coldvoiceclap_attribute:Warm_vs._Cold-0.491-1.06-0.466-1.00-0.0660.400
186durationvoiceclap_attribute:duration15.792+0.7716.509+0.8810.9146.354
187score background qualityvoiceclap_attribute:score_background_quality3.351+0.213.505+0.523.2450.498
188score content enjoymentvoiceclap_attribute:score_content_enjoyment4.757+0.694.818+1.104.6560.147
189score overall qualityvoiceclap_attribute:score_overall_quality2.634-0.002.797+0.392.6350.412
190score speech qualityvoiceclap_attribute:score_speech_quality1.897+0.091.879-0.181.8910.069
191talking speedvoiceclap_attribute:talking_speed4.498-1.214.029-1.339.0283.756
All Orange embedding components

The cosine above is the meaningful vector-level comparison; component values have no standalone semantic label.

Timbre — all 128 vector components
DimensionTinyBaseOrange teacher
00.25570.29860.2585
1-0.1336-0.1128-0.0905
2-0.0985-0.1056-0.0894
3-0.0779-0.0414-0.0346
40.01800.02650.0437
50.05290.03560.0288
6-0.0894-0.0935-0.0826
70.09760.03210.0013
80.14040.09750.1326
90.07160.08130.0684
100.0097-0.0286-0.0226
11-0.1000-0.0853-0.1066
120.0127-0.0152-0.0916
13-0.0971-0.0701-0.0899
14-0.1082-0.0924-0.0948
150.06870.06700.0427
16-0.0957-0.0772-0.0736
170.08750.05640.0787
18-0.0550-0.0488-0.0699
19-0.0808-0.0894-0.0862
200.12690.10820.1198
210.03630.0163-0.0838
22-0.0808-0.0914-0.0857
23-0.0471-0.0513-0.0225
24-0.0332-0.03860.0034
25-0.0764-0.0874-0.0800
26-0.0329-0.0635-0.0511
270.0124-0.0681-0.0850
280.0001-0.0747-0.0481
290.0126-0.0154-0.0268
300.21920.23670.2222
310.03320.11430.0854
320.14130.12950.1365
33-0.1519-0.1371-0.1004
340.0085-0.00330.0369
350.22790.18180.0806
36-0.1115-0.0838-0.0979
37-0.0952-0.1021-0.1078
38-0.1130-0.0940-0.0909
39-0.0942-0.1036-0.0779
400.08650.07620.0081
41-0.1317-0.1163-0.0804
420.05220.01420.0273
43-0.0401-0.0752-0.0938
440.17310.14320.0906
45-0.0351-0.0612-0.0309
46-0.0577-0.0503-0.0189
470.04300.04470.0268
48-0.00420.02270.0154
490.07160.10720.1124
500.09090.07770.0361
51-0.0961-0.0777-0.0196
52-0.0358-0.00250.0623
53-0.1034-0.0919-0.0929
54-0.0913-0.0762-0.0956
55-0.0488-0.0460-0.0876
560.09470.08530.0200
57-0.1163-0.1138-0.0999
580.07640.06700.0584
59-0.00030.0223-0.0314
600.06970.0858-0.0271
610.03320.0356-0.0047
620.12110.13100.2065
630.07350.10460.0673
640.01270.08530.0597
65-0.0584-0.0833-0.0743
660.02780.06370.0576
670.02510.05560.0584
68-0.0572-0.00870.0716
69-0.0538-0.0691-0.0834
700.15090.11990.0865
71-0.0178-0.04110.0085
72-0.1058-0.0995-0.0882
730.02550.0019-0.0397
74-0.1053-0.0924-0.0995
750.03580.0404-0.0028
76-0.0788-0.1183-0.0884
770.11200.10410.2041
780.04810.01500.1223
79-0.0889-0.0726-0.0790
800.03440.07110.1489
81-0.1259-0.1199-0.0825
82-0.1096-0.1214-0.0872
83-0.0440-0.0340-0.0057
84-0.1067-0.0975-0.1129
85-0.0445-0.0612-0.0967
860.06970.07870.1083
87-0.0716-0.0843-0.0887
880.14330.14730.1398
890.08170.0731-0.0035
90-0.0498-0.04240.0314
91-0.0808-0.1041-0.0837
92-0.1101-0.0904-0.0884
930.18360.15850.0689
940.06030.06040.1053
95-0.04350.03330.0653
96-0.1053-0.0592-0.0476
97-0.04180.00240.0569
98-0.0256-0.0899-0.0920
99-0.0437-0.01970.0889
1000.11390.11020.1506
101-0.0711-0.0787-0.0881
102-0.01440.0010-0.0912
103-0.0836-0.0955-0.0961
104-0.0735-0.0660-0.0287
105-0.0531-0.0899-0.0878
106-0.1110-0.1300-0.1095
107-0.0976-0.0742-0.0523
1080.08270.08990.1284
1090.01440.01630.0417
1100.05720.02490.0375
111-0.1178-0.1143-0.1057
112-0.1110-0.1351-0.0898
113-0.0942-0.1321-0.1282
114-0.0423-0.1021-0.0981
115-0.0351-0.0929-0.0467
116-0.0880-0.0772-0.0781
1170.16150.11170.1853
118-0.02020.02460.0431
119-0.0860-0.1041-0.0881
1200.0052-0.0138-0.0333
121-0.0406-0.0526-0.0090
122-0.0073-0.0173-0.0891
1230.00180.05510.1220
1240.04330.0500-0.0450
1250.03440.0061-0.0443
126-0.0510-0.0675-0.0442
1270.09180.08530.1060
Identity — all 250 vector components
DimensionTinyBaseOrange teacher
0-0.0255-0.0259-0.0323
10.11780.12240.1202
20.12770.10760.0630
3-0.0913-0.1004-0.0887
4-0.0565-0.0331-0.0425
50.05080.07380.0619
6-0.1088-0.1008-0.0686
7-0.0293-0.05130.0622
8-0.0338-0.0087-0.0865
90.0150-0.01240.0109
100.01450.00330.0497
110.0098-0.00650.0077
12-0.0813-0.0765-0.0512
13-0.0894-0.0693-0.0413
14-0.0676-0.0304-0.0667
15-0.0847-0.0725-0.0551
160.06030.0295-0.0086
17-0.0634-0.0630-0.0562
18-0.0175-0.0018-0.0476
190.07850.10490.0657
200.0231-0.0012-0.0468
21-0.0572-0.0504-0.0434
22-0.0752-0.0558-0.0485
23-0.0248-0.04460.0246
240.01620.0018-0.0399
25-0.02830.0022-0.0296
260.14280.09630.1737
27-0.0286-0.0504-0.0380
28-0.0575-0.0707-0.0506
29-0.0381-0.05540.0085
300.00850.0293-0.0378
31-0.0228-0.0554-0.0587
320.05860.0538-0.0481
330.01190.0030-0.0176
34-0.0220-0.00850.0270
35-0.0918-0.0869-0.0602
36-0.0690-0.0689-0.0494
37-0.0287-0.0331-0.0323
38-0.0329-0.0369-0.0020
39-0.0657-0.0320-0.0674
40-0.0447-0.0644-0.0542
41-0.0212-0.0059-0.0108
42-0.0653-0.0675-0.0465
430.03620.05670.0714
440.00020.00050.0922
45-0.0556-0.0547-0.0324
46-0.0456-0.05060.0089
470.14570.13320.0851
48-0.0681-0.0513-0.0499
490.06340.07200.0524
500.01310.01050.0130
51-0.01150.00080.0150
52-0.0103-0.0419-0.0376
530.05230.02360.0417
54-0.0724-0.0729-0.0790
550.16080.09950.0548
56-0.0525-0.0232-0.0643
57-0.00230.0230-0.0088
58-0.0395-0.0324-0.0331
59-0.0286-0.0360-0.0651
60-0.0352-0.0371-0.0490
610.0253-0.00040.0886
62-0.0643-0.0556-0.0411
630.04800.0729-0.0508
64-0.0601-0.0511-0.0521
65-0.01570.0165-0.0571
66-0.0534-0.0569-0.0392
67-0.0497-0.0236-0.0492
680.04190.06390.0746
69-0.0111-0.0224-0.0051
70-0.0941-0.0950-0.0678
71-0.0104-0.0155-0.0490
720.00540.0040-0.0030
73-0.0936-0.1103-0.0834
74-0.0257-0.02460.0850
75-0.0195-0.0356-0.0062
760.0193-0.0322-0.0445
77-0.0302-0.02640.0179
780.05200.05630.0020
79-0.0362-0.0358-0.0500
800.09930.13950.1410
810.05390.03170.0536
82-0.0442-0.0358-0.0449
830.04110.0477-0.0415
840.09740.09090.1281
85-0.0463-0.0693-0.0500
86-0.0345-0.0331-0.0446
87-0.0170-0.0308-0.0399
88-0.0615-0.0475-0.0493
89-0.0138-0.0443-0.0452
900.03550.00360.0559
910.01560.0078-0.0587
92-0.0329-0.0441-0.0432
930.08130.08190.1633
94-0.0381-0.0520-0.0759
95-0.0523-0.0698-0.0521
96-0.0471-0.0594-0.0446
97-0.0589-0.0511-0.0113
98-0.0333-0.0797-0.0528
99-0.0305-0.0344-0.0625
100-0.0421-0.0626-0.0441
101-0.0785-0.1053-0.0472
1020.02530.0216-0.0536
103-0.0390-0.05330.0120
104-0.0809-0.0608-0.0651
105-0.0385-0.0198-0.0649
106-0.0572-0.0468-0.0385
1070.00660.01620.0010
108-0.0006-0.0358-0.0371
1090.00730.01340.0867
110-0.0297-0.0199-0.0189
1110.00930.01170.0892
112-0.0307-0.0338-0.0403
113-0.0264-0.05720.0033
114-0.0158-0.03780.1013
115-0.0202-0.0075-0.0354
1160.01030.0005-0.0070
1170.06950.03530.0065
1180.13810.15210.1414
1190.02590.0108-0.0384
1200.05060.01380.0514
121-0.0261-0.0169-0.0537
1220.0002-0.0117-0.0472
1230.0207-0.01460.0110
124-0.0851-0.0932-0.0498
1250.18440.21430.1018
1260.15510.13230.1260
127-0.0188-0.0271-0.0436
128-0.0048-0.0200-0.0468
129-0.0044-0.01330.0250
130-0.0288-0.0248-0.0502
1310.10690.07070.1542
1320.01210.00570.0833
1330.12670.09630.0980
1340.05890.08280.0369
1350.06720.06480.0979
136-0.0142-0.0335-0.0293
1370.08940.0797-0.0228
1380.06950.0909-0.0112
139-0.0056-0.0380-0.0511
140-0.0851-0.0538-0.0589
141-0.00640.0109-0.0462
142-0.0497-0.0493-0.0432
1430.09320.12510.0237
144-0.0629-0.0653-0.0520
1450.01800.0271-0.0249
1460.13530.09950.1443
147-0.0175-0.04750.0236
1480.0078-0.0608-0.0551
149-0.0955-0.1148-0.0643
150-0.0539-0.0549-0.0491
151-0.0610-0.0425-0.0569
152-0.0520-0.0824-0.0554
153-0.0620-0.0657-0.0130
154-0.0262-0.0450-0.0487
155-0.0733-0.0770-0.0651
156-0.0144-0.0071-0.0424
157-0.0215-0.0585-0.0889
158-0.0402-0.0342-0.0384
159-0.00070.0084-0.0584
160-0.0293-0.0127-0.0474
161-0.0648-0.0190-0.0182
162-0.0000-0.0080-0.0565
163-0.1083-0.0824-0.0556
164-0.1007-0.0801-0.0439
1650.10310.09410.1190
166-0.0456-0.0594-0.0581
1670.08560.12420.0850
168-0.0847-0.0999-0.0596
169-0.0648-0.1076-0.0569
170-0.0196-0.0385-0.0467
1710.07990.08730.1047
1720.03590.0203-0.0035
1730.0051-0.0019-0.0359
174-0.0761-0.0603-0.0513
175-0.00480.0077-0.0007
176-0.0409-0.0165-0.0424
177-0.0333-0.0560-0.0411
1780.03520.0522-0.0254
1790.05250.02130.1007
1800.09360.07160.0513
181-0.0435-0.0432-0.0651
182-0.0499-0.0416-0.0431
183-0.0584-0.0837-0.0231
1840.14090.15480.1246
1850.12390.11970.1154
186-0.0743-0.0761-0.0357
1870.07280.05940.0126
1880.04660.04500.0648
189-0.0837-0.0556-0.0792
1900.12580.15390.0436
1910.11920.07970.1852
192-0.0223-0.0249-0.0604
1930.0183-0.0122-0.0526
1940.00850.0143-0.0516
1950.08320.08060.1299
196-0.0714-0.0468-0.0485
1970.11780.08010.0771
198-0.0388-0.0567-0.0482
199-0.0302-0.0215-0.0548
200-0.0681-0.0450-0.0577
2010.01210.0037-0.0211
202-0.0880-0.0981-0.0701
203-0.0433-0.0398-0.0623
2040.0409-0.00240.0326
205-0.0228-0.0513-0.0195
2060.0033-0.0369-0.0515
207-0.00080.03350.0865
2080.03190.02950.0064
2090.11350.08780.1329
2100.12300.0959-0.0074
2110.00740.0120-0.0384
2120.15040.17100.1071
213-0.0343-0.04570.0219
214-0.0079-0.0295-0.0528
2150.01190.02500.0020
216-0.0577-0.0437-0.0598
217-0.1022-0.0891-0.0667
2180.03030.02800.0620
219-0.01940.0172-0.0034
220-0.0244-0.03980.0604
221-0.0264-0.0405-0.0433
2220.07050.10440.0014
2230.02490.0078-0.0280
224-0.0601-0.0621-0.0456
225-0.0055-0.0385-0.0415
226-0.0591-0.0869-0.0304
227-0.0570-0.0241-0.0659
228-0.0174-0.02880.0052
2290.0032-0.0179-0.0453
2300.10120.10130.1578
2310.08040.07880.0911
232-0.0842-0.0446-0.0500
2330.0402-0.00950.0032
2340.05890.00230.0048
2350.15320.14130.1693
236-0.0499-0.0473-0.0428
237-0.0393-0.0576-0.0509
238-0.0459-0.0299-0.0547
2390.11870.14850.1361
2400.04710.03920.0884
2410.01990.0191-0.0007
242-0.0378-0.0183-0.0735
243-0.0847-0.0720-0.0641
2440.04890.0504-0.0392
245-0.0326-0.0468-0.0475
2460.11070.12150.0524
247-0.0615-0.0457-0.0420
248-0.0314-0.0504-0.0017
2490.03930.07160.0724
Source: P3 source-disjoint test split · p3_02009875 · SHA-256 0bae3ca8a72fbb3d…

Reproducibility. The model repository provides the per-target normalization JSON, inference and training code, and the independent test evaluation. P3 test clips are source-disjoint but cherry-picked using teacher scores. Their examples must not be used to estimate average accuracy. Audio provenance is stated under each card.