Song Emotion Recognition on RAVDESS Song
85.8Weighted AccuracyVQ-MAE-S (Frame)
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| VQ-MAE-S (Frame)Upstream=Finetune, Downstream=Query2Emo, Model Category=Specialist Model2023.12 | 85.8 | — | 85.7 | |
| emotion2vecUpstream=Freeze2023.12 | 85 | 85.2 | 84.8 | |
| VQ-MAE-S (Frame)Upstream=Finetune2023.12 | 84.2 | — | 84.3 | |
| VQ-MAE-S (Patch-tf)Upstream=Finetune2023.12 | 84 | — | 84 | |
| VQ-MAE-S (Patch-tf)Upstream=Finetune, Downstream=Query2Emo, Model Category=Specialist Model2023.12 | 83.7 | — | 83.4 | |
| data2vec 2.0Upstream=Freeze2023.12 | 73 | 74.6 | 72.7 | |
| L³-NetUpstream=Freeze, Downstream=Linear2023.12 | 71 | — | — | |
| data2vecUpstream=Freeze2023.12 | 63.8 | 64.1 | 63.4 | |
| WavLM-base+Upstream=Freeze2023.12 | 54.9 | 53.9 | 54.2 | |
| SpecMAEUpstream=Finetune2023.12 | 54.5 | — | 53.9 | |
| WavLM-baseUpstream=Freeze2023.12 | 52.3 | 52.4 | 52.1 |