Audio Classification on Edansa (test)
96.6Anth F1-scoreAST
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| ASTlearning rate (lr)=.0001, batch size (bs)=16, optimizer=Adam, sliding window size=10s, step size=10s2026.05 | 96.6 | 94.9 | 84.9 | 92.1 | 0.117 | |
| SSASTlearning rate (lr)=.00001, batch size (bs)=16, optimizer=Adam, sliding window size=10s, step size=10s2026.05 | 96.2 | 94 | 82.1 | 90.8 | 0.061 | |
| CNN 10learning rate (lr)=.001, batch size (bs)=32, optimizer=Adam, sliding window size=10s, step size=10s2026.05 | 95.4 | 95.6 | 86.5 | 92.5 | 0.071 | |
| ResNet-50learning rate (lr)=.001, batch size (bs)=32, optimizer=Adam, sliding window size=10s, step size=10s2026.05 | 95.1 | 94.8 | 86 | 92 | 0.1 | |
| PaSSTlearning rate (lr)=.0001, batch size (bs)=16, optimizer=AdamW, finetuning protocol=FullFT, sliding window size=10s, step size=10s2026.05 | 94.9 | 93.3 | 84.6 | 90.9 | 0.131 | |
| EfficientNet-B7learning rate (lr)=.001, batch size (bs)=64, optimizer=AdamW, sliding window size=10s, step size=10s2026.05 | 94.8 | 94.7 | 84.7 | 91.4 | 0.23 | |
| Qwen2-Audiolearning rate (lr)=.00001, batch size (bs)=4, optimizer=AdamW, sliding window size=10s, step size=10s2026.05 | 94.5 | 94.7 | 83.2 | 90.8 | 0.756 | |
| CNN 14learning rate (lr)=.0001, batch size (bs)=64, optimizer=AdamW, sliding window size=10s, step size=10s2026.05 | 94.2 | 95.6 | 86.5 | 92.1 | 0.078 | |
| CLAPlearning rate (lr)=.00001, batch size (bs)=16, optimizer=Adam, finetuning protocol=FullFT, sliding window size=10s, step size=10s2026.05 | 94 | 95.4 | 79.5 | 89.7 | 0.195 | |
| Whisperlearning rate (lr)=.00001, batch size (bs)=16, optimizer=AdamW, sliding window size=10s, step size=10s2026.05 | 93.7 | 95 | 83.3 | 90.7 | 0.207 | |
| AVESlearning rate (lr)=.0001, batch size (bs)=16, optimizer=Adam, finetuning protocol=FullFT, sliding window size=10s, step size=10s2026.05 | 93.2 | 93.1 | 83.2 | 89.8 | 0.065 | |
| W2V2learning rate (lr)=.00001, batch size (bs)=16, optimizer=Adam, sliding window size=10s, step size=10s2026.05 | 90.2 | 89.6 | 81.3 | 87 | 0.068 | |
| BirdNETautotune settings=default autotune settings, sliding window size=10s, step size=10s2026.05 | 87.1 | 91.1 | 74.7 | 84.3 | — |