System Detail
iic/eres2net-large-3dspeaker
eres2net_large_3dspeaker
Submitted by: Public checkpoint
- Global position
- #20 / 25
- Scenario-macro EER
- 14.1682
- Scenario-macro minDCF
- 0.593516
- 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:
iic/eres2net-large-3dspeaker - Submitted by: Public checkpoint
- Metadata: OpenSVBench model record
- Source: IIC / ModelScope
- Architecture: ERes2Net
- Model family: ERes2Net
- Submitted alias:
speech_eres2net_large_sv_zh-cn_3dspeaker_16k - Created at:
2026-05-29T14:58:14.607926+00:00
Training And Links
Training data and references
- Training data: 3D-Speaker open dataset, about 10k Chinese speakers
- Training setup: Official 16 kHz Chinese ERes2Net-Large release; the local model card reports 18.3M parameters and training on 3D-Speaker.
- Score processing: None
- Submission mode:
full-core - Model source: Open link
- Paper: An Enhanced Res2Net with Local and Global Feature Fusion for Speaker Verification
Higher-Ranked Scenarios
Scenario ranks above the global position
Show 3 supporting trial sets
- HI-MIA: #3 on its trial-set ranking.
- AliMeeting Near/Far: #3 on its trial-set ranking.
- AliMeeting Overlap: #4 on its trial-set ranking.
Lower-Ranked Scenarios
Scenario ranks below the global position
Show 3 supporting trial sets
- GLOBE: #25 on its trial-set ranking.
- VoxCeleb Short: #25 on its trial-set ranking.
- VoxPopuli Aging: #25 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 |
|---|---|---|---|---|
|
Channel / Device Robustness
How resilient the model is to device and channel mismatch.
|
#5/25
Competitive
|
Device and channel variation
2.3442 EER from leader
|
6.3103
0.401249 minDCF
|
|
|
Distance Robustness
How much performance changes across meeting and domestic distance mismatch.
|
#7/25
Competitive
|
Distance mismatch
4.5696 EER from leader
|
9.2638
0.454491 minDCF
|
|
|
Accent / Dialect Robustness
How reliably the model tracks identity across accent and dialect mismatch.
|
#8/25
Competitive
|
Accent and dialect variation
4.6832 EER from leader
|
20.4447
0.733333 minDCF
|
|
|
Overlap Robustness
How well speaker identity survives light, mid, and heavy overlap in both meeting and domestic recordings.
|
#17/25
Needs work
|
Overlapping speakers
6.6750 EER from leader
|
14.3667
0.552132 minDCF
|
|
|
Genre-Shift Robustness
Whether performance holds when CN-Celeb enrollment and test speech come from different source genres.
|
#19/25
Needs work
|
Source-genre variation
15.2450 EER from leader
|
28.4850
0.888960 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.
|
#20/25
Needs work
|
Noise and reverberation
6.4440 EER from leader
|
8.7040
0.382320 minDCF
|
|
|
Short-Duration Robustness
How much performance holds up when speech evidence is limited by duration.
|
#22/25
Needs work
|
Short-duration speech
8.9763 EER from leader
|
17.5708
0.654520 minDCF
|
|
|
Cross-Lingual Robustness
How well speaker identity survives enrollment-test language mismatch.
|
#23/25
Needs work
|
Language mismatch
6.8960 EER from leader
|
8.5536
0.499376 minDCF
|
|
|
Speaking-Style Robustness
Whether the model can preserve identity across emotion-driven change, whispered speech, and noise-induced Lombard speaking style.
|
#23/25
Needs work
|
Speaking style shift
6.7691 EER from leader
|
8.6857
0.465867 minDCF
|
|
|
In-The-Wild Robustness
How strong the model is on unconstrained celebrity and media speech across official CN-Celeb and VoxCeleb protocols.
|
#23/25
Needs work
|
Open-domain media speech
11.8702 EER from leader
|
14.2563
0.581636 minDCF
|
|
|
Aging Robustness
How stable identity representations remain across age-derived and longitudinal recording gaps.
|
#24/25
Needs work
|
Speaker aging and time gaps
17.4176 EER from leader
|
19.2094
0.914797 minDCF
|
Variant Compare
Compare ERes2Net variants
Compare the family variants on the overall board, a Scenario, or a Trial Set.
Show
Variant Compare
Compare ERes2Net 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 |
|---|---|---|---|---|---|---|---|---|
|
iic/eres2netv2-zh
ERes2NetV2
|
Public checkpoint | IIC | Large Chinese speaker corpus, about 200k speakers | Official 16 kHz Chinese ERes2NetV2 release. | #14 | 11.5449 | 0.484961 | Open |
|
iic/eres2net-en
ERes2Net
|
Public checkpoint | IIC | VoxCeleb2 development set, 5,994 speakers | Official 16 kHz English ERes2Net release. | #16 | 12.0955 | 0.472802 | Open |
|
iic/eres2net-large-3dspeaker
ERes2Net
Current
|
Public checkpoint | IIC / ModelScope | 3D-Speaker open dataset, about 10k Chinese speakers | Official 16 kHz Chinese ERes2Net-Large release; the local model card reports 18.3M parameters and training on 3D-Speaker. | #20 | 14.1682 | 0.593516 | 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
|
#23/25
|
Cross-lingual
|
200000 |
8.5536
0.499376 minDCF
|
|
HI-MIA
HI-MIA
|
#3/25
|
Short-duration
|
660000 |
2.8963
0.178460 minDCF
|
|
GSC Short
Google Speech Commands
|
#19/25
|
Short-duration
|
220000 |
13.3500
0.658090 minDCF
|
|
GLOBE
GLOBE
|
#25/25
|
Accent/dialect
|
56848 |
28.8893
0.801800 minDCF
|
|
CN-Celeb
CN-Celeb
|
#19/25
|
In-the-wild
|
3484292 |
14.0524
0.486731 minDCF
|
|
CN-Celeb Genre
CN-Celeb
|
#19/25
|
Genre shift
|
440000 |
28.4850
0.888960 minDCF
|
|
CN-Celeb Short
CN-Celeb
|
#18/25
|
Short-duration
|
546964 |
26.8023
0.858622 minDCF
|
|
VoxCeleb1-O
VoxCeleb1
|
#25/25
|
In-the-wild
|
37611 |
13.0837
0.638176 minDCF
|
|
VoxCeleb1-E
VoxCeleb1
|
#25/25
|
In-the-wild
|
579818 |
13.1915
0.655802 minDCF
|
|
VoxCeleb1-H
VoxCeleb1
|
#25/25
|
In-the-wild
|
550894 |
17.1053
0.735645 minDCF
|
|
VoxCeleb Short
VoxCeleb1
|
#25/25
|
Short-duration
|
394724 |
27.2344
0.922910 minDCF
|
|
3D-Speaker Device
3D-Speaker
|
#4/25
|
Channel/device
|
180000 |
6.8733
0.451193 minDCF
|
|
3D-Speaker Distance
3D-Speaker
|
#6/25
|
Distance
|
175163 |
10.3440
0.537387 minDCF
|
|
3D-Speaker Dialect
3D-Speaker
|
#5/25
|
Accent/dialect
|
180000 |
12.0000
0.664867 minDCF
|
|
FFSVC 2022 Cross-Channel
FFSVC 2022
|
#7/25
|
Channel/device
|
72000 |
5.7472
0.351306 minDCF
|
|
FFSVC 2022 Cross-Domain
FFSVC 2022
|
#7/25
|
Distance
|
66546 |
5.8125
0.357548 minDCF
|
|
Whisper40 Whisper
Whisper40
|
#21/25
|
Speaking style
|
17600 |
14.8750
0.868437 minDCF
|
|
Lombard Grid Lombard
Lombard Grid
|
#21/25
|
Speaking style
|
29524 |
1.7139
0.100335 minDCF
|
|
VOiCES Noise/Reverb
VOiCES
|
#20/25
|
Noise/reverb
|
55000 |
8.7040
0.382320 minDCF
|
|
CHiME-6 Domestic Far-Field
CHiME-6
|
#20/25
|
Distance
|
36487 |
17.2988
0.733554 minDCF
|
|
CHiME-6 Overlap
CHiME-6
|
#22/25
|
Overlap
|
39600 |
23.8333
0.823583 minDCF
|
|
ESD
ESD
|
#22/25
|
Speaking style
|
437408 |
9.4684
0.428828 minDCF
|
|
AliMeeting Near/Far
AliMeeting
|
#3/25
|
Distance
|
220000 |
3.6000
0.189475 minDCF
|
|
AliMeeting Overlap
AliMeeting
|
#4/25
|
Overlap
|
165000 |
4.9000
0.280680 minDCF
|
|
VoxKnesset
VoxKnesset
|
#23/25
|
Aging
|
158312 |
20.2049
0.837278 minDCF
|
|
VoxPopuli Aging
VoxPopuli
|
#25/25
|
Aging
|
146575 |
18.2139
0.992315 minDCF
|