Trial Set
ESD
Builds cross-emotion SV trials from speaker and emotion labels.
- Source dataset
- ESD
- Trials
- 437408
- Protocol source
- Self-constructed protocol
- Source split
- Public ESD recordings
Source Dataset
About ESD
The Emotional Speech Database (ESD) is a controlled bilingual corpus created for emotional voice conversion and related speech research. It contains parallel sentences from 10 native English speakers and 10 native Chinese speakers, spoken in five emotion categories, with more than 29 hours of recordings. OpenSVBench uses speaker and emotion labels to construct its own cross-emotion verification trials.
- Languages
- English and Mandarin Chinese
- Speakers
- 20 (10 per language)
- Emotion categories
- Neutral, happy, angry, sad, and surprise
- Scale
- 350 parallel sentences; 29+ hours
Direct Ranking
Systems on this trial set
23 reviewed systems currently have valid results.
| Rank | System | Submitted by | Trial-Set EER ↓ | Trial-Set minDCF ↓ |
|---|---|---|---|---|
| #1 |
AntSpeaker/MECT-B2
antspeaker-mect-b2-vb2-full-core
|
Public checkpoint | 1.2675 | 0.096087 |
| #2 |
Dual-LoRA W2V-BERT2-Adapter-MFA (TidyVoice)
duallora-w2vbert2-tidyvoice-full-core
|
shangguanqituan | 1.7750 | 0.135385 |
| #3 |
W2V-BERT 2.0 + MHFA (VoxCeleb2 + VoxBlink2)
Wav2Vec2-BERT
|
pengjy | 1.7975 | 0.148715 |
| #4 |
palabraai/redimnet2-b6-vb2+vox2
redimnet2-b6-vb2vox2-lm-full-core
|
Public checkpoint | 2.3127 | 0.164424 |
| #5 |
wespeaker/w2vbert2
vb2-w2vbert
|
Public checkpoint | 2.5000 | 0.175313 |
| #6 |
Agora Voiceprint (ResNet293)
agora_vp
|
agora_vp | 2.5025 | 0.150464 |
| #7 |
wespeaker/samresnet100
vb2-sam100
|
Public checkpoint | 2.7503 | 0.223690 |
| #8 |
wespeaker/res152-voxceleb
wespeaker_resnet152
|
Public checkpoint | 3.0742 | 0.225959 |
| #9 |
palabraai/redimnet2-b6
redimnet2
|
Public checkpoint | 3.0850 | 0.221939 |
| #10 |
wespeaker/res293-voxceleb
wespeaker_resnet293
|
Public checkpoint | 3.2714 | 0.243165 |
| #11 |
wespeaker/res34-voxceleb
wespeaker_resnet34_voxceleb
|
Public checkpoint | 3.5211 | 0.254288 |
| #12 |
iic/eres2net-en
eres_en
|
Public checkpoint | 4.3650 | 0.310919 |
| #13 |
wespeaker/ecapa512-voxceleb
wespeaker_ecapa512
|
Public checkpoint | 4.4900 | 0.305853 |
| #14 |
wespeaker/ecapa1024-voxceleb
wespeaker_ecapa1024
|
Public checkpoint | 4.6018 | 0.330355 |
| #15 |
wespeaker/campplus-voxceleb
campplus_wespeaker
|
Public checkpoint | 4.6625 | 0.346034 |
| #16 |
speechbrain/ecapa
ecapa
|
Public checkpoint | 5.7925 | 0.396047 |
| #17 |
iic/eres2netv2-zh
eresv2_zh
|
Public checkpoint | 7.5779 | 0.346889 |
| #18 |
wespeaker/res34-cnceleb
wespeaker_r34
|
Public checkpoint | 8.0200 | 0.419287 |
| #19 |
iic/eres2net-large-3dspeaker
eres2net_large_3dspeaker
|
Public checkpoint | 9.4684 | 0.428828 |
| #20 |
iic/campplus
campplus
|
Public checkpoint | 9.6571 | 0.470165 |
| #21 |
microsoft/unispeech-sat-base-plus-sv
unispeech_sat_base_plus
|
Public checkpoint | 12.8825 | 0.923664 |
| #22 |
speechbrain/xvector
xvector
|
Public checkpoint | 13.0389 | 0.693274 |
| #23 |
microsoft/wavlm-base-plus-sv
wavlm_base
|
Public checkpoint | 13.3674 | 0.882281 |