OpenSVBench Scenario-driven SV leaderboard
System Detail

wespeaker/ res34-cnceleb

wespeaker_r34

ranked review: approved WeSpeaker ResNet

This page shows the global rank, scenario ranks, and trial-set results for this submission.

Global Position

#14/18

Public ranking position among reviewed full-core submissions.

Scenario-Macro EER
14.4910
Scenario-Macro minDCF
0.607964
Coverage

26/26

0 scenarios ranked #1 1 scenario in top 3
Public Listing

Ranked Publicly

This full-core result is approved and included in the public ranking.

Model Provenance

Source and architecture

  • Displayed name: wespeaker/res34-cnceleb
  • Source: WeSpeaker
  • Architecture: ResNet
  • Submitted alias: wespeaker-cnceleb-resnet34-LM
  • Created at: 2026-05-29T14:26:58.126430+00:00
Training And Links

Training data and references

  • Training data: CN-Celeb train
  • Training setup: ResNet34 r-vector with TSTP pooling and large-margin fine-tuning on the CN-Celeb WeSpeaker recipe.
  • Submission mode: full-core
  • Paper: Deep Residual Learning for Image Recognition
Higher-Ranked Scenarios

Scenario ranks above the global position

Show 3 supporting trial sets
  • CN-Celeb Short: #1 on its trial-set ranking.
  • CN-Celeb Genre: #2 on its trial-set ranking.
  • CN-Celeb: #3 on its trial-set ranking.
Lower-Ranked Scenarios

Scenario ranks below the global position

Show 3 supporting trial sets
  • VoxPopuli Aging: #16 on its trial-set ranking.
  • Lombard Grid Lombard: #16 on its trial-set ranking.
  • 3D-Speaker Device: #15 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
Genre-Shift Robustness
Whether performance holds when CN-Celeb enrollment and test speech come from different source genres.
#2/18 Top 3
Source-genre variation 1.4230 EER from leader
15.6955
0.591628 minDCF
Short-Duration Robustness
How much performance holds up when speech evidence is limited by duration.
#7/18 Competitive
Short-duration speech 1.7754 EER from leader
14.2898
0.579413 minDCF
In-The-Wild Robustness
How strong the model is on unconstrained celebrity and media speech across official CN-Celeb and VoxCeleb protocols.
#10/18 Needs work
Open-domain media speech 2.6600 EER from leader
7.8434
0.404734 minDCF
Speaking-Style Robustness
Whether the model can preserve identity across emotion-driven change, whispered speech, and noise-induced Lombard speaking style.
#11/18 Needs work
Speaking style shift 4.2849 EER from leader
7.0743
0.410279 minDCF
Cross-Lingual Robustness
How well speaker identity survives enrollment-test language mismatch.
#13/18 Needs work
Language mismatch 3.1094 EER from leader
7.5770
0.460771 minDCF
Aging Robustness
How stable identity representations remain across age-derived and longitudinal recording gaps.
#13/18 Needs work
Speaker aging and time gaps 10.0589 EER from leader
12.1285
0.635341 minDCF
Overlap Robustness
How well speaker identity survives light, mid, and heavy overlap in both meeting and domestic recordings.
#14/18 Needs work
Overlapping speakers 6.2739 EER from leader
16.4122
0.671276 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.
#14/18 Needs work
Noise and reverberation 14.0600 EER from leader
16.6400
0.531160 minDCF
Accent / Dialect Robustness
How reliably the model tracks identity across accent and dialect mismatch.
#14/18 Needs work
Accent and dialect variation 5.3952 EER from leader
25.8303
0.825746 minDCF
Distance Robustness
How much performance changes across meeting and domestic distance mismatch.
#15/18 Needs work
Distance mismatch 7.7286 EER from leader
16.9924
0.739294 minDCF
Channel / Device Robustness
How resilient the model is to device and channel mismatch.
#15/18 Needs work
Device and channel variation 12.6075 EER from leader
18.9178
0.837962 minDCF
Variant Compare

Compare ResNet variants

Expand this section to compare checkpoints in the same architecture group by source, training data, training setup, and leaderboard result.

Show
Variant Source Training Data Training Setup Global Rank Macro EER Macro minDCF Open
wespeaker/res152-voxceleb WeSpeaker VoxCeleb2 dev, 5,994 speakers ResNet152-TSTP-emb256 r-vector, ArcMargin, 150-epoch VoxCeleb2 recipe with speed perturbation and MUSAN/RIRS augmentation. #4
ranked
11.2275 0.442748 Open
wespeaker/res293-voxceleb WeSpeaker VoxCeleb2 dev, 5,994 speakers ResNet293-TSTP-emb256 r-vector, large-margin fine-tuned; official card reports 28.62M parameters and 28.10G FLOPs. #5
ranked
11.3130 0.439607 Open
wespeaker/res34-voxceleb WeSpeaker VoxCeleb2 dev, 5,994 speakers ResNet34-TSTP-emb256 r-vector, large-margin fine-tuned; official card reports 6.63M parameters and 4.55G FLOPs. #7
ranked
12.0516 0.477054 Open
wespeaker/res34-cnceleb
Current
WeSpeaker CN-Celeb train ResNet34 r-vector with TSTP pooling and large-margin fine-tuning on the CN-Celeb WeSpeaker recipe. #14
ranked
14.4910 0.607964 Open
Trial-Set Results

Raw ranking by trial set

Expand this section to inspect the detailed trial-set rankings.

Show
Trial Set Rank Scenarios Trials Score
TidyVoiceX2-ASV
TidyVoiceX2-ASV
#13/18
Cross-lingual
200000
7.5770
0.460771 minDCF
HI-MIA
HI-MIA
#12/18
Short-duration
660000
7.7542
0.426640 minDCF
GSC Short
Google Speech Commands
#4/18
Short-duration
220000
8.3650
0.496165 minDCF
GLOBE
GLOBE
#12/18
Accent/dialect
56848
26.3293
0.658746 minDCF
CN-Celeb
CN-Celeb
#3/18
In-the-wild
3484292
7.4729
0.345407 minDCF
CN-Celeb Genre
CN-Celeb
#2/18
Genre shift
440000
15.6955
0.591628 minDCF
CN-Celeb Short
CN-Celeb
#1/18
Short-duration
546964
17.6675
0.600899 minDCF
VoxCeleb1-O
VoxCeleb1
#15/18
In-the-wild
37611
7.0897
0.433848 minDCF
VoxCeleb1-E
VoxCeleb1
#15/18
In-the-wild
579818
7.1685
0.432123 minDCF
VoxCeleb1-H
VoxCeleb1
#15/18
In-the-wild
550894
10.3836
0.526212 minDCF
VoxCeleb Short
VoxCeleb1
#14/18
Short-duration
394724
23.3725
0.793947 minDCF
3D-Speaker Device
3D-Speaker
#15/18
Channel/device
180000
26.8967
0.998813 minDCF
3D-Speaker Distance
3D-Speaker
#14/18
Distance
175163
25.1560
0.990655 minDCF
3D-Speaker Dialect
3D-Speaker
#14/18
Accent/dialect
180000
25.3313
0.992747 minDCF
FFSVC 2022 Cross-Channel
FFSVC 2022
#9/18
Channel/device
72000
10.9389
0.677111 minDCF
FFSVC 2022 Cross-Domain
FFSVC 2022
#9/18
Distance
66546
11.0514
0.689180 minDCF
Whisper40 Whisper
Whisper40
#5/18
Speaking style
17600
9.8125
0.615313 minDCF
Lombard Grid Lombard
Lombard Grid
#16/18
Speaking style
29524
3.3905
0.196237 minDCF
VOiCES Noise/Reverb
VOiCES
#14/18
Noise/reverb
55000
16.6400
0.531160 minDCF
CHiME-6 Domestic Far-Field
CHiME-6
#14/18
Distance
36487
22.9123
0.914742 minDCF
CHiME-6 Overlap
CHiME-6
#13/18
Overlap
39600
23.6778
0.833472 minDCF
ESD
ESD
#13/18
Speaking style
437408
8.0200
0.419287 minDCF
AliMeeting Near/Far
AliMeeting
#7/18
Distance
220000
8.8500
0.362600 minDCF
AliMeeting Overlap
AliMeeting
#7/18
Overlap
165000
9.1467
0.509080 minDCF
VoxKnesset
VoxKnesset
#13/18
Aging
158312
14.3733
0.632640 minDCF
VoxPopuli Aging
VoxPopuli
#16/18
Aging
146575
9.8837
0.638041 minDCF