Trial Set
VOiCES Noise/Reverb
Uses the 6,400-recording devkit test split, not the VOiCES 2019 challenge trial keys.
Notice: This is not the original official SV trial list.
- Source dataset
- VOiCES
- Trials
- 55000
- Protocol source
- Self-constructed protocol subset
- Source split
- VOiCES devkit test split
Source Dataset
About VOiCES
VOiCES (Voices Obscured in Complex Environmental Settings) was recorded in furnished rooms to capture far-field speech with natural room acoustics and distractor sound. Clean foreground speech from LibriSpeech was played alongside noise such as television, music, or babble and captured by microphones placed around the rooms. OpenSVBench uses the devkit test split and builds its own noise/reverberation trials.
- Source speech
- LibriSpeech clean utterances
- Recording
- Furnished rooms; 12 distributed microphones
- Acoustic conditions
- Room reverberation and TV, music, or babble noise
- Original use
- Far-field speech and speaker-recognition research
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 |
palabraai/redimnet2-b6-vb2+vox2
redimnet2-b6-vb2vox2-lm-full-core
|
Public checkpoint | 2.2600 | 0.073780 |
| #2 |
W2V-BERT 2.0 + MHFA (VoxCeleb2 + VoxBlink2)
Wav2Vec2-BERT
|
pengjy | 2.4000 | 0.086400 |
| #3 |
wespeaker/w2vbert2
vb2-w2vbert
|
Public checkpoint | 2.5800 | 0.074040 |
| #4 |
AntSpeaker/MECT-B2
antspeaker-mect-b2-vb2-full-core
|
Public checkpoint | 2.6540 | 0.087120 |
| #5 |
wespeaker/samresnet100
vb2-sam100
|
Public checkpoint | 2.9800 | 0.098660 |
| #6 |
Dual-LoRA W2V-BERT2-Adapter-MFA (TidyVoice)
duallora-w2vbert2-tidyvoice-full-core
|
shangguanqituan | 3.0800 | 0.084760 |
| #7 |
palabraai/redimnet2-b6
redimnet2
|
Public checkpoint | 3.1200 | 0.114220 |
| #8 |
iic/eres2net-en
eres_en
|
Public checkpoint | 4.7200 | 0.148640 |
| #9 |
wespeaker/res152-voxceleb
wespeaker_resnet152
|
Public checkpoint | 5.2600 | 0.154640 |
| #10 |
wespeaker/res293-voxceleb
wespeaker_resnet293
|
Public checkpoint | 5.3600 | 0.148780 |
| #11 |
speechbrain/ecapa
ecapa
|
Public checkpoint | 5.5800 | 0.213320 |
| #12 |
wespeaker/res34-voxceleb
wespeaker_resnet34_voxceleb
|
Public checkpoint | 5.9280 | 0.181180 |
| #13 |
Agora Voiceprint (ResNet293)
agora_vp
|
agora_vp | 6.4000 | 0.171760 |
| #14 |
wespeaker/campplus-voxceleb
campplus_wespeaker
|
Public checkpoint | 6.4600 | 0.220820 |
| #15 |
wespeaker/ecapa1024-voxceleb
wespeaker_ecapa1024
|
Public checkpoint | 6.6060 | 0.214720 |
| #16 |
wespeaker/ecapa512-voxceleb
wespeaker_ecapa512
|
Public checkpoint | 7.6160 | 0.264660 |
| #17 |
iic/eres2net-large-3dspeaker
eres2net_large_3dspeaker
|
Public checkpoint | 8.7040 | 0.382320 |
| #18 |
iic/eres2netv2-zh
eresv2_zh
|
Public checkpoint | 11.6800 | 0.402500 |
| #19 |
wespeaker/res34-cnceleb
wespeaker_r34
|
Public checkpoint | 16.6400 | 0.531160 |
| #20 |
speechbrain/xvector
xvector
|
Public checkpoint | 19.0800 | 0.568120 |
| #21 |
microsoft/unispeech-sat-base-plus-sv
unispeech_sat_base_plus
|
Public checkpoint | 20.9800 | 0.741000 |
| #22 |
microsoft/wavlm-base-plus-sv
wavlm_base
|
Public checkpoint | 21.0600 | 0.742060 |
| #23 |
iic/campplus
campplus
|
Public checkpoint | 23.4800 | 0.737920 |