Emotion Recognition on ER
70.62AccuracyWavLM Large
Evaluation results
| Method | Links | |
|---|---|---|
| WavLM Large# Params=316.6M, Evaluation Protocol=Frozen-encoder2026.06 | 70.62 | |
| WavLM Base+# Params=94.7M, Evaluation Protocol=Frozen-encoder2026.06 | 68.65 | |
| HuBERT Large# Params=316.6M, Evaluation Protocol=Frozen-encoder2026.06 | 67.62 | |
| BEATS_iter2#Param=90M, Pre-training Data=AS2022.12 | 66.1 | |
| WavLM Base# Params=94.7M, Evaluation Protocol=Frozen-encoder2026.06 | 65.94 | |
| BEATS_iter1#Param=90M, Pre-training Data=AS2022.12 | 65.9 | |
| wav2vec 2.0 Large# Params=316.6M, Evaluation Protocol=Frozen-encoder2026.06 | 65.64 | |
| BEATS_iter3+#Param=90M, Pre-training Data=AS2022.12 | 65 | |
| HuBERT Base# Params=94.7M, Evaluation Protocol=Frozen-encoder2026.06 | 64.92 | |
| HuBERTEvaluation protocol=Frozen setting2021.10 | 64.9 | |
| HuBERTfrozen setting=true2024.05 | 64.9 | |
| S-JEPA# Params=51.8M, Evaluation Protocol=Frozen-encoder2026.06 | 64.83 | |
| BEATS_iter3#Param=90M, Pre-training Data=AS2022.12 | 64.5 | |
| wav2vec 2.0 Base# Params=95.0M, Evaluation Protocol=Frozen-encoder2026.06 | 63.43 | |
| Wav2vec 2.0Evaluation protocol=Frozen setting2021.10 | 63.4 | |
| Wav2vec 2.0#Param=95M, Pre-training Data=LS, Pre-training Type=Self-Supervised Pre-Training, Evaluation Protocol=SUPERB frozen2022.12 | 63.4 | |
| Wav2Vec 2.0frozen setting=true2024.05 | 63.4 | |
| DistilHuBERT# Params=23.5M, Evaluation Protocol=Frozen-encoder2026.06 | 63.02 | |
| DeCoAR 2.0# Params=89.8M, Evaluation Protocol=Frozen-encoder2026.06 | 62.47 | |
| PANN#Param=81M, Pre-training Type=No Pre-Training2022.12 | 61.8 | |
| SSAMBA-base2024.05 | 61.5 | |
| SSAMBAModel size=base, Pretrained=true, Head learning rate=normal2024.05 | 61.5 | |
| modified CPC# Params=1.8M, Evaluation Protocol=Frozen-encoder2026.06 | 60.96 | |
| SSAST-FramePre-training data=AudioSet + Librispeech, Patch type=Frame-based2021.10 | 60.5 | |
| Wav2vecPre-training framework=Discriminative2021.10 | 59.8 | |
| Wav2vec#Param=33M, Pre-training Data=LS, Pre-training Type=Self-Supervised Pre-Training2022.12 | 59.8 | |
| MAE-AST#Param=86M, Pre-training Data=AS+LS, Pre-training Type=Self-Supervised Pre-Training2022.12 | 59.8 | |
| Wav2Vec2024.05 | 59.8 | |
| wav2vec# Params=32.5M, Evaluation Protocol=Frozen-encoder2026.06 | 59.79 | |
| VQ-APC# Params=4.6M, Evaluation Protocol=Frozen-encoder2026.06 | 59.66 | |
| SSAST-PatchPre-training data=AudioSet + Librispeech, Patch type=Patch-based2021.10 | 59.6 | |
| SS-AST#Param=89M, Pre-training Data=AS+LS, Pre-training Type=Self-Supervised Pre-Training2022.12 | 59.6 | |
| SSASTModel size=base, Pretrained=true, Head learning rate=normal2024.05 | 59.6 | |
| APC# Params=4.1M, Evaluation Protocol=Frozen-encoder2026.06 | 59.33 | |
| APCPre-training framework=Generative/Autoregressive2021.10 | 59.3 | |
| APC2024.05 | 59.3 | |
| NPC# Params=19.4M, Evaluation Protocol=Frozen-encoder2026.06 | 59.08 | |
| SSASTModel size=small, Pretrained=true, Head learning rate=larger2024.05 | 58.7 | |
| SSAST-PatchPre-training data=Librispeech Only, Patch type=Patch-based2021.10 | 58.3 | |
| vq-wav2vec# Params=34.2M, Evaluation Protocol=Frozen-encoder2026.06 | 58.24 | |
| SSAMBA-small2024.05 | 58.2 | |
| SSAMBAModel size=small, Pretrained=true, Head learning rate=larger2024.05 | 58.2 | |
| SSAMBAModel size=base, Pretrained=false, Head learning rate=normal2024.05 | 58.2 | |
| PASE+# Params=7.8M, Evaluation Protocol=Frozen-encoder2026.06 | 57.86 | |
| SSAMBAModel size=small, Pretrained=false, Head learning rate=larger2024.05 | 57.1 | |
| SSAMBA-tiny2024.05 | 56.3 | |
| SSAMBAModel size=tiny, Pretrained=true, Head learning rate=larger2024.05 | 56.3 | |
| SSAMBAModel size=tiny, Pretrained=false, Head learning rate=larger2024.05 | 56.3 | |
| TERA# Params=21.3M, Evaluation Protocol=Frozen-encoder2026.06 | 56.27 | |
| AST#Param=86M, Pre-training Data=IN, Pre-training Type=Out-of-domain Supervised Pre-Training2022.12 | 56 | |
| SSASTModel size=tiny, Pretrained=true, Head learning rate=larger2024.05 | 55.7 | |
| Mockingjay# Params=85.1M, Evaluation Protocol=Frozen-encoder2026.06 | 50.28 |