Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

VoxCeleb

Benchmarks

Task NameDataset NameSOTA ResultTrend
Speaker RecognitionVoxCeleb1 (test)
EER2.21
126
Speaker VerificationVoxCeleb1 (test)
Cosine EER0.856
80
Speaker IdentificationVoxCeleb1
Accuracy97.5
58
Speaker VerificationVoxCeleb1 Hard Cleaned
EER0.0099
45
Speaker VerificationVoxCeleb1 Cleaned (Extended)
EER (%)0.48
45
Speaker VerificationVoxCeleb1 (Vox1-O)
EER0.627
33
Speaker VerificationVoxCeleb1 extended (test)
EER1.07
25
Speaker verificationVoxCeleb 1 (verification)
EER0.48
22
Speaker VerificationVoxCeleb1 (Vox1-H)
EER0.986
20
Face ReenactmentVoxCeleb2 (test)
FID24.92
16
Face ReenactmentVoxCeleb1 (test)
SSIM0.804
16
Speaker VerificationVoxCeleb Hard 1
EER (f-f)1.87
15
Speaker VerificationVoxCeleb Extended 1
EER (f-f)1
15
Speaker VerificationVoxCeleb-E
EER (f-f)0.97
15
Cross-modal verificationVoxCeleb1 (Unseen-Unheard)
AUC85
13
Speech SeparationVoxCeleb2-2Mix (test)
SDRi13.1
12
Speaker RecognitionVoxCeleb1 extended (vox1-e)
EER (mean)0.9
11
Speaker RecognitionVoxCeleb1 original (vox1-o)
EER (mean)0.74
11
ImperceptibilityVoxCeleb2
SSIM0.961
10
Speaker VerificationVoxCeleb 1hr context Normal
EER0.0094
10
Speaker VerificationVoxCeleb 10min context Normal
EER1.04
10
Neural Field ReconstructionVoxCeleb2
PSNR (Step 1)29.84
9
Video self-reconstructionVoxceleb1 (test)
L1 Loss0.0354
9
Cross-identity face animationVoxceleb 1
ARD2.399
9
Cross-modal verificationVoxCeleb1 (Seen-Heard)
AUC0.937
9
Showing 25 of 88 rows