Trial Set
Whisper40 Whisper
Builds same-text neutral-to-whisper SV trials because no upstream SV trial keys are used.
- Source dataset
- Whisper40
- Trials
- 17600
- Protocol source
- Self-constructed protocol
- Source split
- Provided Whisper40 archive
Source Dataset
About Whisper40
Whisper40 is a Mandarin corpus recorded to study speaker recognition across neutral and whispered speech. Forty native speakers each read matched text in both speaking styles, yielding 3,200 utterances and about 656 minutes of audio. OpenSVBench constructs same-text neutral-to-whisper trials from the released recordings.
- Language
- Mandarin Chinese
- Speakers
- 40 (21 male, 19 female)
- Recordings
- 3,200 utterances; about 656 minutes
- Conditions
- Matched neutral and whispered speech
Direct Ranking
Systems on this trial set
23 reviewed systems currently have valid results.
| Rank | System | Submitted by | Trial-Set EER ↓ | Trial-Set minDCF ↓ |
|---|---|---|---|---|
| #1 |
W2V-BERT 2.0 + MHFA (VoxCeleb2 + VoxBlink2)
Wav2Vec2-BERT
|
pengjy | 3.9375 | 0.310375 |
| #2 |
palabraai/redimnet2-b6-vb2+vox2
redimnet2-b6-vb2vox2-lm-full-core
|
Public checkpoint | 4.7875 | 0.269813 |
| #3 |
Dual-LoRA W2V-BERT2-Adapter-MFA (TidyVoice)
duallora-w2vbert2-tidyvoice-full-core
|
shangguanqituan | 5.2562 | 0.320875 |
| #4 |
wespeaker/w2vbert2
vb2-w2vbert
|
Public checkpoint | 5.8125 | 0.291250 |
| #5 |
AntSpeaker/MECT-B2
antspeaker-mect-b2-vb2-full-core
|
Public checkpoint | 7.5625 | 0.354313 |
| #6 |
iic/eres2netv2-zh
eresv2_zh
|
Public checkpoint | 9.1250 | 0.663000 |
| #7 |
wespeaker/res152-voxceleb
wespeaker_resnet152
|
Public checkpoint | 9.2500 | 0.567375 |
| #8 |
Agora Voiceprint (ResNet293)
agora_vp
|
agora_vp | 9.3750 | 0.516687 |
| #9 |
wespeaker/res293-voxceleb
wespeaker_resnet293
|
Public checkpoint | 9.5562 | 0.533062 |
| #10 |
wespeaker/res34-cnceleb
wespeaker_r34
|
Public checkpoint | 9.8125 | 0.615313 |
| #11 |
wespeaker/samresnet100
vb2-sam100
|
Public checkpoint | 10.1250 | 0.485938 |
| #12 |
palabraai/redimnet2-b6
redimnet2
|
Public checkpoint | 10.1875 | 0.646312 |
| #13 |
iic/campplus
campplus
|
Public checkpoint | 11.2625 | 0.683063 |
| #14 |
wespeaker/res34-voxceleb
wespeaker_resnet34_voxceleb
|
Public checkpoint | 11.4375 | 0.645688 |
| #15 |
iic/eres2net-en
eres_en
|
Public checkpoint | 11.8750 | 0.666250 |
| #16 |
wespeaker/campplus-voxceleb
campplus_wespeaker
|
Public checkpoint | 14.0000 | 0.689750 |
| #17 |
wespeaker/ecapa1024-voxceleb
wespeaker_ecapa1024
|
Public checkpoint | 14.8000 | 0.736062 |
| #18 |
iic/eres2net-large-3dspeaker
eres2net_large_3dspeaker
|
Public checkpoint | 14.8750 | 0.868437 |
| #19 |
speechbrain/ecapa
ecapa
|
Public checkpoint | 15.6250 | 0.849062 |
| #20 |
wespeaker/ecapa512-voxceleb
wespeaker_ecapa512
|
Public checkpoint | 16.6250 | 0.780062 |
| #21 |
speechbrain/xvector
xvector
|
Public checkpoint | 32.1250 | 0.999063 |
| #22 |
microsoft/wavlm-base-plus-sv
wavlm_base
|
Public checkpoint | 32.6563 | 0.989187 |
| #23 |
microsoft/unispeech-sat-base-plus-sv
unispeech_sat_base_plus
|
Public checkpoint | 33.6875 | 0.997938 |