System Detail
wespeaker/campplus-voxceleb
campplus_wespeaker
Submitted by: Public checkpoint
- Global position
- #17 / 25
- Scenario-macro EER
- 13.4165
- Scenario-macro minDCF
- 0.520675
- Trial-set coverage
- 26 / 26
- Scenario top-three
- 0
Publicly ranked
This result is included in the reviewed public ranking.
Model Provenance
Source and architecture
- Displayed name:
wespeaker/campplus-voxceleb - Submitted by: Public checkpoint
- Metadata: OpenSVBench model record
- Source: WeSpeaker
- Architecture: CAMPPlus
- Model family: CAMPPlus
- Submitted alias:
wespeaker-voxceleb-campplus-LM - Created at:
2026-05-29T14:20:28.222436+00:00
Training And Links
Training data and references
- Training data: VoxCeleb2 dev
- Training setup: CAMPPlus with TSTP pooling, ArcMargin large-margin fine-tuning, and MUSAN/RIRS augmentation.
- Score processing: None
- Submission mode:
full-core - Model source: Open link
- Paper: CAM++: A Fast and Efficient Network for Speaker Verification Using Context-Aware Masking
Higher-Ranked Scenarios
Scenario ranks above the global position
Show 3 supporting trial sets
- VoxCeleb1-O: #13 on its trial-set ranking.
- VoxCeleb1-E: #14 on its trial-set ranking.
- VoxCeleb Short: #14 on its trial-set ranking.
Lower-Ranked Scenarios
Scenario ranks below the global position
Show 3 supporting trial sets
- CN-Celeb Genre: #21 on its trial-set ranking.
- GSC Short: #21 on its trial-set ranking.
- CN-Celeb: #21 on its trial-set ranking.
Scenario Rankings
Full ranking across scenarios
This table shows where the model sits on each scenario, using source-balanced scenario scores and the linked trial sets as evidence.
| Scenario | Rank | Lens | Evidence | Score |
|---|---|---|---|---|
|
Aging Robustness
How stable identity representations remain across age-derived and longitudinal recording gaps.
|
#15/25
Needs work
|
Speaker aging and time gaps
1.8152 EER from leader
|
3.6069
0.160596 minDCF
|
|
|
Overlap Robustness
How well speaker identity survives light, mid, and heavy overlap in both meeting and domestic recordings.
|
#16/25
Needs work
|
Overlapping speakers
6.3899 EER from leader
|
14.0816
0.622128 minDCF
|
|
|
Cross-Lingual Robustness
How well speaker identity survives enrollment-test language mismatch.
|
#17/25
Needs work
|
Language mismatch
4.5274 EER from leader
|
6.1851
0.403050 minDCF
|
|
|
Speaking-Style Robustness
Whether the model can preserve identity across emotion-driven change, whispered speech, and noise-induced Lombard speaking style.
|
#17/25
Needs work
|
Speaking style shift
4.4408 EER from leader
|
6.3574
0.358438 minDCF
|
|
|
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.
|
#17/25
Needs work
|
Noise and reverberation
4.2000 EER from leader
|
6.4600
0.220820 minDCF
|
|
|
Distance Robustness
How much performance changes across meeting and domestic distance mismatch.
|
#19/25
Needs work
|
Distance mismatch
9.9849 EER from leader
|
14.6792
0.634129 minDCF
|
|
|
Short-Duration Robustness
How much performance holds up when speech evidence is limited by duration.
|
#19/25
Needs work
|
Short-duration speech
7.7549 EER from leader
|
16.3494
0.638979 minDCF
|
|
|
Accent / Dialect Robustness
How reliably the model tracks identity across accent and dialect mismatch.
|
#19/25
Needs work
|
Accent and dialect variation
8.1668 EER from leader
|
23.9283
0.721471 minDCF
|
|
|
In-The-Wild Robustness
How strong the model is on unconstrained celebrity and media speech across official CN-Celeb and VoxCeleb protocols.
|
#20/25
Needs work
|
Open-domain media speech
6.0176 EER from leader
|
8.4036
0.311533 minDCF
|
|
|
Channel / Device Robustness
How resilient the model is to device and channel mismatch.
|
#20/25
Needs work
|
Device and channel variation
13.9139 EER from leader
|
17.8800
0.826833 minDCF
|
|
|
Genre-Shift Robustness
Whether performance holds when CN-Celeb enrollment and test speech come from different source genres.
|
#21/25
Needs work
|
Source-genre variation
16.4097 EER from leader
|
29.6497
0.829445 minDCF
|
Variant Compare
Compare CAMPPlus variants
Compare the family variants on the overall board, a Scenario, or a Trial Set.
Show
Variant Compare
Compare CAMPPlus variants
Compare the family variants on the overall board, a Scenario, or a Trial Set.
| Variant | Submitted by | Source | Training Data | Training Setup | Rank | EER | minDCF | Open |
|---|---|---|---|---|---|---|---|---|
|
wespeaker/campplus-voxceleb
CAMPPlus
Current
|
Public checkpoint | WeSpeaker | VoxCeleb2 dev | CAMPPlus with TSTP pooling, ArcMargin large-margin fine-tuning, and MUSAN/RIRS augmentation. | #17 | 13.4165 | 0.520675 | Open |
|
iic/campplus
CAMPPlus
|
Public checkpoint | IIC | Large Chinese speaker corpus, about 200k speakers | Official ModelScope CAMPPlus release. | #23 | 15.1677 | 0.590178 | Open |
Trial-Set Results
Raw ranking by trial set
Expand this section to inspect the detailed trial-set rankings.
Show
Raw ranking by trial set
Expand this section to inspect the detailed trial-set rankings.
| Trial Set | Rank | Scenarios | Trials | Score |
|---|---|---|---|---|
|
TidyVoiceX2-ASV
TidyVoiceX2-ASV
|
#17/25
|
Cross-lingual
|
200000 |
6.1851
0.403050 minDCF
|
|
HI-MIA
HI-MIA
|
#20/25
|
Short-duration
|
660000 |
7.9797
0.395343 minDCF
|
|
GSC Short
Google Speech Commands
|
#21/25
|
Short-duration
|
220000 |
16.4345
0.764210 minDCF
|
|
GLOBE
GLOBE
|
#18/25
|
Accent/dialect
|
56848 |
25.6966
0.530070 minDCF
|
|
CN-Celeb
CN-Celeb
|
#21/25
|
In-the-wild
|
3484292 |
15.7139
0.551833 minDCF
|
|
CN-Celeb Genre
CN-Celeb
|
#21/25
|
Genre shift
|
440000 |
29.6497
0.829445 minDCF
|
|
CN-Celeb Short
CN-Celeb
|
#20/25
|
Short-duration
|
546964 |
27.4153
0.854569 minDCF
|
|
VoxCeleb1-O
VoxCeleb1
|
#13/25
|
In-the-wild
|
37611 |
0.7180
0.061049 minDCF
|
|
VoxCeleb1-E
VoxCeleb1
|
#14/25
|
In-the-wild
|
579818 |
0.8789
0.054292 minDCF
|
|
VoxCeleb1-H
VoxCeleb1
|
#15/25
|
In-the-wild
|
550894 |
1.6832
0.098356 minDCF
|
|
VoxCeleb Short
VoxCeleb1
|
#14/25
|
Short-duration
|
394724 |
13.5682
0.541793 minDCF
|
|
3D-Speaker Device
3D-Speaker
|
#20/25
|
Channel/device
|
180000 |
23.4600
0.938000 minDCF
|
|
3D-Speaker Distance
3D-Speaker
|
#19/25
|
Distance
|
175163 |
22.2628
0.902270 minDCF
|
|
3D-Speaker Dialect
3D-Speaker
|
#19/25
|
Accent/dialect
|
180000 |
22.1600
0.912873 minDCF
|
|
FFSVC 2022 Cross-Channel
FFSVC 2022
|
#21/25
|
Channel/device
|
72000 |
12.3000
0.715667 minDCF
|
|
FFSVC 2022 Cross-Domain
FFSVC 2022
|
#21/25
|
Distance
|
66546 |
12.4278
0.723857 minDCF
|
|
Whisper40 Whisper
Whisper40
|
#19/25
|
Speaking style
|
17600 |
14.0000
0.689750 minDCF
|
|
Lombard Grid Lombard
Lombard Grid
|
#19/25
|
Speaking style
|
29524 |
0.4098
0.039531 minDCF
|
|
VOiCES Noise/Reverb
VOiCES
|
#17/25
|
Noise/reverb
|
55000 |
6.4600
0.220820 minDCF
|
|
CHiME-6 Domestic Far-Field
CHiME-6
|
#16/25
|
Distance
|
36487 |
13.0540
0.549864 minDCF
|
|
CHiME-6 Overlap
CHiME-6
|
#14/25
|
Overlap
|
39600 |
16.9444
0.668250 minDCF
|
|
ESD
ESD
|
#18/25
|
Speaking style
|
437408 |
4.6625
0.346034 minDCF
|
|
AliMeeting Near/Far
AliMeeting
|
#21/25
|
Distance
|
220000 |
10.9720
0.360525 minDCF
|
|
AliMeeting Overlap
AliMeeting
|
#20/25
|
Overlap
|
165000 |
11.2187
0.576007 minDCF
|
|
VoxKnesset
VoxKnesset
|
#15/25
|
Aging
|
158312 |
5.0067
0.213777 minDCF
|
|
VoxPopuli Aging
VoxPopuli
|
#17/25
|
Aging
|
146575 |
2.2071
0.107415 minDCF
|