OpenSVBench speaker verification leaderboard.
Global ranking compares reviewed full-core submissions. Scenario pages group related evaluation conditions. Trial-set pages show the raw ranked results behind each scenario.
Dataset names are shown as source metadata inside the scenario and trial-set views.
Current #1 model: wespeaker/w2vbert2 with 9.9604 scenario-macro EER.
Ranked Top 5
- Global ranking: reviewed full-core submissions ranked by source-balanced scenario-macro EER.
- Scenarios: high-level evaluation conditions.
- Trial sets: raw ranked benchmark entries.
- Datasets: source metadata shown inside scenario and trial-set views.
Ranking
Every ranked row is maintainer-reviewed and recomputed against the current benchmark definition, so rankings stay comparable as the leaderboard is updated.
wespeaker/w2vbert2
Lowest scenario-macro EER across the current benchmark.
wespeaker/w2vbert2
Highest count of top-three high-level scenario ranks on the current board.
| Rank | Model | Global Score | Higher-Ranked Scenarios | Lower-Ranked Scenarios |
|---|---|---|---|---|
| #1 |
wespeaker/w2vbert2
vb2-w2vbert
|
9.9604
scenario-macro EER
0.360767 scenario-macro minDCF
|
||
| #2 |
palabraai/redimnet2-b6
redimnet2
|
10.5277
scenario-macro EER
0.419432 scenario-macro minDCF
|
||
| #3 |
wespeaker/samresnet100
vb2-sam100
|
10.5780
scenario-macro EER
0.390239 scenario-macro minDCF
|
||
| #4 |
wespeaker/res152-voxceleb
wespeaker_resnet152
|
11.2275
scenario-macro EER
0.442748 scenario-macro minDCF
|
||
| #5 |
wespeaker/res293-voxceleb
wespeaker_resnet293
|
11.3130
scenario-macro EER
0.439607 scenario-macro minDCF
|
||
| #6 |
iic/eres2netv2-zh
eresv2_zh
|
11.5449
scenario-macro EER
0.484961 scenario-macro minDCF
|
||
| #7 |
wespeaker/res34-voxceleb
wespeaker_resnet34_voxceleb
|
12.0516
scenario-macro EER
0.477054 scenario-macro minDCF
|
||
| #8 |
iic/eres2net-en
eres_en
|
12.0955
scenario-macro EER
0.472802 scenario-macro minDCF
|
||
| #9 |
wespeaker/campplus-voxceleb
campplus_wespeaker
|
13.4165
scenario-macro EER
0.520675 scenario-macro minDCF
|
||
| #10 |
speechbrain/ecapa
ecapa
|
13.5461
scenario-macro EER
0.543660 scenario-macro minDCF
|
||
| #11 |
wespeaker/ecapa1024-voxceleb
wespeaker_ecapa1024
|
14.0835
scenario-macro EER
0.534111 scenario-macro minDCF
|
||
| #12 |
iic/eres2net-large-3dspeaker
eres2net_large_3dspeaker
|
14.1682
scenario-macro EER
0.593516 scenario-macro minDCF
|
||
| #13 |
wespeaker/ecapa512-voxceleb
wespeaker_ecapa512
|
14.3120
scenario-macro EER
0.541141 scenario-macro minDCF
|
||
| #14 |
wespeaker/res34-cnceleb
wespeaker_r34
|
14.4910
scenario-macro EER
0.607964 scenario-macro minDCF
|
||
| #15 |
iic/campplus
campplus
|
15.1677
scenario-macro EER
0.590178 scenario-macro minDCF
|
||
| #16 |
speechbrain/xvector
xvector
|
23.3539
scenario-macro EER
0.776078 scenario-macro minDCF
|
||
| #17 |
microsoft/wavlm-base-plus-sv
wavlm_base
|
25.1577
scenario-macro EER
0.875090 scenario-macro minDCF
|
||
| #18 |
microsoft/unispeech-sat-base-plus-sv
unispeech_sat_base_plus
|
25.6711
scenario-macro EER
0.890736 scenario-macro minDCF
|
Scenarios covered by OpenSVBench
Each scenario describes a high-level evaluation condition, while the underlying trial sets and datasets remain visible.
Accent / Dialect Robustness
How reliably the model tracks identity across accent and dialect mismatch.
Aging Robustness
How stable identity representations remain across age-derived and longitudinal recording gaps.
Channel / Device Robustness
How resilient the model is to device and channel mismatch.
Cross-Lingual Robustness
How well speaker identity survives enrollment-test language mismatch.
Distance Robustness
How much performance changes across meeting and domestic distance mismatch.
Genre-Shift Robustness
Whether performance holds when CN-Celeb enrollment and test speech come from different source genres.
In-The-Wild Robustness
How strong the model is on unconstrained celebrity and media speech across official CN-Celeb and VoxCeleb protocols.
Noise / Reverb Robustness
How reliably the model preserves identity when room acoustics, distractor noise, and microphone placement deviate from an easier in-corpus reference condition.
Overlap Robustness
How well speaker identity survives light, mid, and heavy overlap in both meeting and domestic recordings.
Short-Duration Robustness
How much performance holds up when speech evidence is limited by duration.
Speaking-Style Robustness
Whether the model can preserve identity across emotion-driven change, whispered speech, and noise-induced Lombard speaking style.