System Detail
iic/eres2net-en
eres_en
Submitted by: Public checkpoint
- Global position
- #13 / 23
- Scenario-macro EER
- 12.0955
- Scenario-macro minDCF
- 0.472802
- 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-en - Submitted by: Public checkpoint
- Metadata: OpenSVBench model record
- Source: IIC
- Architecture: ERes2Net
- Model family: ERes2Net
- Submitted alias:
speech_eres2net_sv_en_voxceleb_16k - Created at:
2026-05-29T14:32:30.802353+00:00
Training And Links
Training data and references
- Training data: VoxCeleb2 development set, 5,994 speakers
- Training setup: Official 16 kHz English ERes2Net release.
- 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
- VOiCES Noise/Reverb: #8 on its trial-set ranking.
- GLOBE: #10 on its trial-set ranking.
- VoxPopuli Aging: #11 on its trial-set ranking.
Lower-Ranked Scenarios
Scenario ranks below the global position
Show 3 supporting trial sets
- GSC Short: #17 on its trial-set ranking.
- CN-Celeb Short: #16 on its trial-set ranking.
- FFSVC 2022 Cross-Domain: #16 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 |
|---|---|---|---|---|
|
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.
|
#8/23
Competitive
|
Noise and reverberation
2.4600 EER from leader
|
4.7200
0.148640 minDCF
|
|
|
Overlap Robustness
How well speaker identity survives light, mid, and heavy overlap in both meeting and domestic recordings.
|
#11/23
Competitive
|
Overlapping speakers
5.6546 EER from leader
|
13.7639
0.580907 minDCF
|
|
|
Accent / Dialect Robustness
How reliably the model tracks identity across accent and dialect mismatch.
|
#11/23
Competitive
|
Accent and dialect variation
5.0934 EER from leader
|
21.9151
0.690782 minDCF
|
|
|
Speaking-Style Robustness
Whether the model can preserve identity across emotion-driven change, whispered speech, and noise-induced Lombard speaking style.
|
#12/23
Competitive
|
Speaking style shift
3.5712 EER from leader
|
5.4878
0.330691 minDCF
|
|
|
Cross-Lingual Robustness
How well speaker identity survives enrollment-test language mismatch.
|
#12/23
Competitive
|
Language mismatch
3.8708 EER from leader
|
5.5284
0.374254 minDCF
|
|
|
Aging Robustness
How stable identity representations remain across age-derived and longitudinal recording gaps.
|
#13/23
Needs work
|
Speaker aging and time gaps
1.8505 EER from leader
|
3.6422
0.155018 minDCF
|
|
|
In-The-Wild Robustness
How strong the model is on unconstrained celebrity and media speech across official CN-Celeb and VoxCeleb protocols.
|
#13/23
Needs work
|
Open-domain media speech
2.9456 EER from leader
|
6.6673
0.251235 minDCF
|
|
|
Distance Robustness
How much performance changes across meeting and domestic distance mismatch.
|
#13/23
Needs work
|
Distance mismatch
7.3264 EER from leader
|
12.5832
0.554135 minDCF
|
|
|
Channel / Device Robustness
How resilient the model is to device and channel mismatch.
|
#14/23
Needs work
|
Device and channel variation
11.1849 EER from leader
|
15.1872
0.730291 minDCF
|
|
|
Short-Duration Robustness
How much performance holds up when speech evidence is limited by duration.
|
#15/23
Needs work
|
Short-duration speech
6.7813 EER from leader
|
15.8159
0.582696 minDCF
|
|
|
Genre-Shift Robustness
Whether performance holds when CN-Celeb enrollment and test speech come from different source genres.
|
#15/23
Needs work
|
Source-genre variation
13.4670 EER from leader
|
27.7395
0.802170 minDCF
|
Variant Compare
Compare ERes2Net variants
Expand this section to compare models in the same model family by architecture, source, training data, training setup, and leaderboard result.
Show
Compare ERes2Net variants
Expand this section to compare models in the same model family by architecture, source, training data, training setup, and leaderboard result.
| Variant | Submitted by | Source | Training Data | Training Setup | Global Rank | Macro EER | Macro minDCF | Open |
|---|---|---|---|---|---|---|---|---|
|
iic/eres2netv2-zh
ERes2NetV2
|
Public checkpoint | IIC | Large Chinese speaker corpus, about 200k speakers | Official 16 kHz Chinese ERes2NetV2 release. |
#11
ranked
|
11.5449 | 0.484961 | Open |
|
iic/eres2net-en
ERes2Net
Current
|
Public checkpoint | IIC | VoxCeleb2 development set, 5,994 speakers | Official 16 kHz English ERes2Net release. |
#13
ranked
|
12.0955 | 0.472802 | Open |
|
iic/eres2net-large-3dspeaker
ERes2Net
|
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. |
#17
ranked
|
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
|
#12/23
|
Cross-lingual
|
200000 |
5.5284
0.374254 minDCF
|
|
HI-MIA
HI-MIA
|
#14/23
|
Short-duration
|
660000 |
7.0967
0.323343 minDCF
|
|
GSC Short
Google Speech Commands
|
#17/23
|
Short-duration
|
220000 |
13.8900
0.693170 minDCF
|
|
GLOBE
GLOBE
|
#10/23
|
Accent/dialect
|
56848 |
25.1935
0.535991 minDCF
|
|
CN-Celeb
CN-Celeb
|
#13/23
|
In-the-wild
|
3484292 |
12.1487
0.429169 minDCF
|
|
CN-Celeb Genre
CN-Celeb
|
#15/23
|
Genre shift
|
440000 |
27.7395
0.802170 minDCF
|
|
CN-Celeb Short
CN-Celeb
|
#16/23
|
Short-duration
|
546964 |
27.2323
0.818788 minDCF
|
|
VoxCeleb1-O
VoxCeleb1
|
#12/23
|
In-the-wild
|
37611 |
0.8028
0.053980 minDCF
|
|
VoxCeleb1-E
VoxCeleb1
|
#14/23
|
In-the-wild
|
579818 |
0.9682
0.061025 minDCF
|
|
VoxCeleb1-H
VoxCeleb1
|
#14/23
|
In-the-wild
|
550894 |
1.7866
0.104900 minDCF
|
|
VoxCeleb Short
VoxCeleb1
|
#14/23
|
Short-duration
|
394724 |
15.0446
0.495483 minDCF
|
|
3D-Speaker Device
3D-Speaker
|
#12/23
|
Channel/device
|
180000 |
18.9467
0.845527 minDCF
|
|
3D-Speaker Distance
3D-Speaker
|
#11/23
|
Distance
|
175163 |
17.8119
0.808851 minDCF
|
|
3D-Speaker Dialect
3D-Speaker
|
#11/23
|
Accent/dialect
|
180000 |
18.6367
0.845573 minDCF
|
|
FFSVC 2022 Cross-Channel
FFSVC 2022
|
#16/23
|
Channel/device
|
72000 |
11.4278
0.615056 minDCF
|
|
FFSVC 2022 Cross-Domain
FFSVC 2022
|
#16/23
|
Distance
|
66546 |
11.5618
0.628141 minDCF
|
|
Whisper40 Whisper
Whisper40
|
#15/23
|
Speaking style
|
17600 |
11.8750
0.666250 minDCF
|
|
Lombard Grid Lombard
Lombard Grid
|
#13/23
|
Speaking style
|
29524 |
0.2235
0.014903 minDCF
|
|
VOiCES Noise/Reverb
VOiCES
|
#8/23
|
Noise/reverb
|
55000 |
4.7200
0.148640 minDCF
|
|
CHiME-6 Domestic Far-Field
CHiME-6
|
#11/23
|
Distance
|
36487 |
12.0892
0.487278 minDCF
|
|
CHiME-6 Overlap
CHiME-6
|
#12/23
|
Overlap
|
39600 |
17.1944
0.658028 minDCF
|
|
ESD
ESD
|
#12/23
|
Speaking style
|
437408 |
4.3650
0.310919 minDCF
|
|
AliMeeting Near/Far
AliMeeting
|
#13/23
|
Distance
|
220000 |
8.8700
0.292270 minDCF
|
|
AliMeeting Overlap
AliMeeting
|
#16/23
|
Overlap
|
165000 |
10.3333
0.503787 minDCF
|
|
VoxKnesset
VoxKnesset
|
#13/23
|
Aging
|
158312 |
5.4308
0.235319 minDCF
|
|
VoxPopuli Aging
VoxPopuli
|
#11/23
|
Aging
|
146575 |
1.8537
0.074717 minDCF
|