System Detail

MECT-b2-czur-finetuned

MECT-b2-czur-finetuned

Submitted by: czur

ranked review: approved mect MECT-B2 MECT
Global position
#5 / 26
Scenario-macro EER
7.2276
Scenario-macro minDCF
0.280633
Trial-set coverage
26 / 26
Scenario top-three
2
Publicly ranked This result is included in the reviewed public ranking.
Model Provenance

Source and architecture

  • Displayed name: MECT-b2-czur-finetuned
  • Submitted by: czur
  • Metadata: Maintainer override
  • Source: mect
  • Architecture: MECT-B2
  • Model family: MECT
  • Submitted alias: MECT-b2-czur-finetuned
  • Created at: 2026-11-02T03:34:20.685023+00:00
Training And Links

Training data and references

  • Training data: VoxCeleb2 dev + VoxBlink2 + 3D-speaker
  • Training setup: Follow AntSpeaker: MECT-B2 adopts 80-dim mean-normalized log-Mel filterbanks (16 kHz, 25 ms window, 10 ms hop, 20–7600 Hz). Its 4-stage CNN-Transformer has C=32, M=3, d=64. It employs frame-level 4-expert MoE, weighted dense fusion, attentive stats pooling and 192-dim projection, trained via SphereFace2. It is further fine-tuned on 3D-Speaker to optimize for dialect variations and varying recording distances.
  • Score processing: None
  • Submission mode: full-core
  • Model source: Open link
  • Paper: MECT: Mixture of Experts with CNN-Transformer Network for Speaker Verification
Higher-Ranked Scenarios

Scenario ranks above the global position

Show 3 supporting trial sets
  • HI-MIA: #1 on its trial-set ranking.
  • 3D-Speaker Dialect: #2 on its trial-set ranking.
  • FFSVC 2022 Cross-Channel: #3 on its trial-set ranking.
Lower-Ranked Scenarios

Scenario ranks below the global position

Show 3 supporting trial sets
  • VoxCeleb1-E: #19 on its trial-set ranking.
  • VoxCeleb1-O: #19 on its trial-set ranking.
  • VoxCeleb1-H: #17 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
Overlap Robustness
How well speaker identity survives light, mid, and heavy overlap in both meeting and domestic recordings.
#2/26 Top 3
Overlapping speakers 0.7054 EER from leader
8.3971
0.410954 minDCF
Channel / Device Robustness
How resilient the model is to device and channel mismatch.
#3/26 Top 3
Device and channel variation 0.9833 EER from leader
3.9661
0.230757 minDCF
Cross-Lingual Robustness
How well speaker identity survives enrollment-test language mismatch.
#4/26 Competitive
Language mismatch 0.2201 EER from leader
1.8679
0.126853 minDCF
Accent / Dialect Robustness
How reliably the model tracks identity across accent and dialect mismatch.
#4/26 Competitive
Accent and dialect variation 0.9365 EER from leader
15.9351
0.447792 minDCF
In-The-Wild Robustness
How strong the model is on unconstrained celebrity and media speech across official CN-Celeb and VoxCeleb protocols.
#5/26 Competitive
Open-domain media speech 1.4269 EER from leader
3.4420
0.154836 minDCF
Distance Robustness
How much performance changes across meeting and domestic distance mismatch.
#5/26 Competitive
Distance mismatch 1.2547 EER from leader
5.2728
0.253762 minDCF
Short-Duration Robustness
How much performance holds up when speech evidence is limited by duration.
#6/26 Competitive
Short-duration speech 3.5852 EER from leader
9.9000
0.379431 minDCF
Genre-Shift Robustness
Whether performance holds when CN-Celeb enrollment and test speech come from different source genres.
#8/26 Competitive
Source-genre variation 11.0067 EER from leader
19.6400
0.649620 minDCF
Speaking-Style Robustness
Whether the model can preserve identity across emotion-driven change, whispered speech, and noise-induced Lombard speaking style.
#9/26 Competitive
Speaking style shift 1.2715 EER from leader
3.0190
0.171062 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.
#12/26 Competitive
Noise and reverberation 2.8220 EER from leader
5.0820
0.139280 minDCF
Aging Robustness
How stable identity representations remain across age-derived and longitudinal recording gaps.
#14/26 Needs work
Speaker aging and time gaps 1.1294 EER from leader
2.9811
0.122619 minDCF
Variant Compare

Compare MECT variants

Compare the family variants on the overall board, a Scenario, or a Trial Set.

Show
Variant Submitted by Source Training Data Training Setup Rank EER minDCF Open
MECT-b2-czur-finetuned
MECT-B2
Current
czur mect VoxCeleb2 dev + VoxBlink2 + 3D-speaker Follow AntSpeaker: MECT-B2 adopts 80-dim mean-normalized log-Mel filterbanks (16 kHz, 25 ms window, 10 ms hop, 20–7600 Hz). Its 4-stage CNN-Transformer has C=32, M=3, d=64. It employs frame-level 4-expert MoE, weighted dense fusion, attentive stats pooling and 192-dim projection, trained via SphereFace2. It is further fine-tuned on 3D-Speaker to optimize for dialect variations and varying recording distances. #5 7.2276 0.280633 Open
AntSpeaker/MECT-B2
MECT-B2
Public checkpoint AntSpeaker VoxCeleb2 dev + VoxBlink2 MECT-B2 uses 80-dimensional mean-normalized log-Mel filterbanks at 16 kHz with a 25 ms window, 10 ms shift, and 20–7600 Hz band, followed by a four-stage CNN-Transformer with C=32, M=3, and Transformer width d=64. The model uses frame-level dense mixture-of-experts with four experts, weighted dense fusion, attentive statistics pooling, and a 192-dimensional projection trained with SphereFace2. The VoxCeleb2 + VoxBlink2 setting uses full training followed by large-margin fine-tuning without speed perturbation; the LMF stage uses VoxCeleb2. #7 8.5599 0.321518 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
#4/26
Cross-lingual
200000
1.8679
0.126853 minDCF
HI-MIA
HI-MIA
#1/26
Short-duration
660000
2.3383
0.120418 minDCF
GSC Short
Google Speech Commands
#9/26
Short-duration
220000
6.3860
0.353380 minDCF
GLOBE
GLOBE
#14/26
Accent/dialect
56848
25.1935
0.495685 minDCF
CN-Celeb
CN-Celeb
#6/26
In-the-wild
3484292
5.6266
0.231593 minDCF
CN-Celeb Genre
CN-Celeb
#8/26
Genre shift
440000
19.6400
0.649620 minDCF
CN-Celeb Short
CN-Celeb
#6/26
Short-duration
546964
18.2769
0.594276 minDCF
VoxCeleb1-O
VoxCeleb1
#19/26
In-the-wild
37611
0.9095
0.066421 minDCF
VoxCeleb1-E
VoxCeleb1
#19/26
In-the-wild
579818
1.1318
0.070483 minDCF
VoxCeleb1-H
VoxCeleb1
#17/26
In-the-wild
550894
1.7309
0.097331 minDCF
VoxCeleb Short
VoxCeleb1
#14/26
Short-duration
394724
12.5989
0.449649 minDCF
3D-Speaker Device
3D-Speaker
#4/26
Channel/device
180000
4.6767
0.273987 minDCF
3D-Speaker Distance
3D-Speaker
#4/26
Distance
175163
5.5433
0.321723 minDCF
3D-Speaker Dialect
3D-Speaker
#2/26
Accent/dialect
180000
6.6767
0.399900 minDCF
FFSVC 2022 Cross-Channel
FFSVC 2022
#3/26
Channel/device
72000
3.2556
0.187528 minDCF
FFSVC 2022 Cross-Domain
FFSVC 2022
#3/26
Distance
66546
3.3137
0.191582 minDCF
Whisper40 Whisper
Whisper40
#8/26
Speaking style
17600
6.8750
0.401000 minDCF
Lombard Grid Lombard
Lombard Grid
#9/26
Speaking style
29524
0.0596
0.003279 minDCF
VOiCES Noise/Reverb
VOiCES
#12/26
Noise/reverb
55000
5.0820
0.139280 minDCF
CHiME-6 Domestic Far-Field
CHiME-6
#10/26
Distance
36487
7.9891
0.352608 minDCF
CHiME-6 Overlap
CHiME-6
#8/26
Overlap
39600
12.7222
0.594389 minDCF
ESD
ESD
#7/26
Speaking style
437408
2.1225
0.108907 minDCF
AliMeeting Near/Far
AliMeeting
#6/26
Distance
220000
4.2450
0.149135 minDCF
AliMeeting Overlap
AliMeeting
#3/26
Overlap
165000
4.0720
0.227520 minDCF
VoxKnesset
VoxKnesset
#14/26
Aging
158312
4.2962
0.199745 minDCF
VoxPopuli Aging
VoxPopuli
#12/26
Aging
146575
1.6660
0.045493 minDCF