Scene-based Audio Classification on HEAR Environmental Sound tasks
78.9ESC-50 AccuracyBYOL-A*
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| BYOL-A*Backbone=AudioNTT2022.09 | 78.9 | 48.9 | 87.5 | 71.8 | — | |
| ABTBackbone=AudioNTT2022.09 | 78.6 | 47.4 | 90.5 | 72.1 | 24 | |
| ABTBackbone=ViTC-B, Patch Size=16x82022.09 | 70.5 | 44.6 | 84.5 | 66.6 | 49 | |
| wav2vec2.0Backbone=wav2vec2.02022.09 | 56.1 | 34.2 | 84.8 | 58.4 | — | |
| CREPEBackbone=CREPE2022.09 | 30.1 | 15.9 | 86.3 | 44.1 | — |