Trial Set
AliMeeting Near/Far
Builds near/far distance trials from public test-set segment metadata.
- Source dataset
- AliMeeting
- Trials
- 220000
- Protocol source
- Self-constructed protocol
- Source split
- Public AliMeeting test-set segments
Source Dataset
About AliMeeting
AliMeeting is a Mandarin corpus of naturally occurring multi-party meetings, released with the M2MeT challenge. Recordings include far-field audio from an eight-channel microphone array as well as near-field headset audio; sessions contain two to four speakers and varying overlap. OpenSVBench uses public test-set segments to construct its own near/far and overlap trials.
- Language
- Mandarin Chinese
- Scale
- About 119 hours of meeting speech
- Recording
- Eight-channel array and headset microphones
- Sessions
- Two to four speakers, with varying overlap
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 |
Agora Voiceprint (ResNet293)
agora_vp
|
agora_vp | 3.3030 | 0.168440 |
| #2 |
iic/eres2net-large-3dspeaker
eres2net_large_3dspeaker
|
Public checkpoint | 3.6000 | 0.189475 |
| #3 |
W2V-BERT 2.0 + MHFA (VoxCeleb2 + VoxBlink2)
Wav2Vec2-BERT
|
pengjy | 5.4400 | 0.180015 |
| #4 |
iic/eres2netv2-zh
eresv2_zh
|
Public checkpoint | 5.8680 | 0.225080 |
| #5 |
wespeaker/w2vbert2
vb2-w2vbert
|
Public checkpoint | 6.6900 | 0.193680 |
| #6 |
Dual-LoRA W2V-BERT2-Adapter-MFA (TidyVoice)
duallora-w2vbert2-tidyvoice-full-core
|
shangguanqituan | 7.2850 | 0.232455 |
| #7 |
AntSpeaker/MECT-B2
antspeaker-mect-b2-vb2-full-core
|
Public checkpoint | 7.7800 | 0.189490 |
| #8 |
palabraai/redimnet2-b6-vb2+vox2
redimnet2-b6-vb2vox2-lm-full-core
|
Public checkpoint | 8.3550 | 0.185310 |
| #9 |
wespeaker/res152-voxceleb
wespeaker_resnet152
|
Public checkpoint | 8.5600 | 0.270260 |
| #10 |
wespeaker/samresnet100
vb2-sam100
|
Public checkpoint | 8.5605 | 0.219880 |
| #11 |
wespeaker/res293-voxceleb
wespeaker_resnet293
|
Public checkpoint | 8.7900 | 0.277705 |
| #12 |
wespeaker/res34-cnceleb
wespeaker_r34
|
Public checkpoint | 8.8500 | 0.362600 |
| #13 |
iic/eres2net-en
eres_en
|
Public checkpoint | 8.8700 | 0.292270 |
| #14 |
wespeaker/res34-voxceleb
wespeaker_resnet34_voxceleb
|
Public checkpoint | 8.8800 | 0.295085 |
| #15 |
palabraai/redimnet2-b6
redimnet2
|
Public checkpoint | 9.3150 | 0.241885 |
| #16 |
iic/campplus
campplus
|
Public checkpoint | 9.3825 | 0.321260 |
| #17 |
speechbrain/ecapa
ecapa
|
Public checkpoint | 10.8485 | 0.362360 |
| #18 |
wespeaker/campplus-voxceleb
campplus_wespeaker
|
Public checkpoint | 10.9720 | 0.360525 |
| #19 |
wespeaker/ecapa1024-voxceleb
wespeaker_ecapa1024
|
Public checkpoint | 11.2950 | 0.346655 |
| #20 |
wespeaker/ecapa512-voxceleb
wespeaker_ecapa512
|
Public checkpoint | 11.4500 | 0.379935 |
| #21 |
speechbrain/xvector
xvector
|
Public checkpoint | 15.6150 | 0.787755 |
| #22 |
microsoft/unispeech-sat-base-plus-sv
unispeech_sat_base_plus
|
Public checkpoint | 23.7800 | 0.959010 |
| #23 |
microsoft/wavlm-base-plus-sv
wavlm_base
|
Public checkpoint | 23.9055 | 0.921520 |