Audio Classification on ESC50 (test)
0.9803R@1 AccuracyMUKA
Evaluation Results
| Method | Links | |
|---|---|---|
| MUKATraining Approach=Training-free, Shots=16, Backbone=PENGI2026.02 | 0.9803 | |
| Linear ProbingTraining Approach=Training-based, Shots=16, Backbone=PENGI2026.02 | 0.96 | |
| PaLMTraining Approach=Training-based, Shots=16, Backbone=PENGI2026.02 | 0.9593 | |
| Treff-AdapterTraining Approach=Training-free, Shots=16, Backbone=PENGI2026.02 | 0.9448 | |
| CoCoOpTraining Approach=Training-based, Shots=16, Context Tokens=16, Backbone=PENGI2026.02 | 0.9427 | |
| CoOpTraining Approach=Training-based, Shots=16, Context Tokens=16, Backbone=PENGI2026.02 | 0.9382 | |
| LAIONZero-shot=true2023.05 | 0.91 | |
| CLAP (ours)Zero-shot=true2023.05 | 0.89 | |
| CLAPZero-shot=true2023.05 | 0.826 | |
| CyCLAPPre-training Dataset=AudioCaps, Zero-shot=true2023.10 | 0.805 | |
| CLAPPre-training Dataset=AudioCaps, Zero-shot=true2023.10 | 0.7925 | |
| CyCLAPsPre-training Dataset=AudioCaps, Zero-shot=true2023.10 | 0.785 | |
| CyCLAPPre-training Dataset=Clotho, Zero-shot=true2023.10 | 0.78 | |
| CyCLAPsPre-training Dataset=Clotho, Zero-shot=true2023.10 | 0.7775 | |
| CyCLAPPre-training Dataset=AudioCaps, Zero-shot=true2023.10 | 0.775 | |
| CyCLAPPre-training Dataset=Clotho, Zero-shot=true2023.10 | 0.77 | |
| CyCLAPSPre-training Dataset=Clotho, Zero-shot=true2023.10 | 0.7575 | |
| CyCLAPSPre-training Dataset=AudioCaps, Zero-shot=true2023.10 | 0.755 | |
| CLAPsPre-training Dataset=AudioCaps, Zero-shot=true2023.10 | 0.75 | |
| CLAPPre-training Dataset=AudioCaps, Zero-shot=true2023.10 | 0.745 | |
| CLAPsPre-training Dataset=Clotho, Zero-shot=true2023.10 | 0.7325 | |
| CLAPPre-training Dataset=Clotho, Zero-shot=true2023.10 | 0.73 | |
| CLAPSPre-training Dataset=AudioCaps, Zero-shot=true2023.10 | 0.7275 | |
| CLAPPre-training Dataset=Clotho, Zero-shot=true2023.10 | 0.7225 | |
| CLAPSPre-training Dataset=Clotho, Zero-shot=true2023.10 | 0.7175 | |
| AudioCLIPZero-shot=true2023.05 | 0.694 | |
| Zero-ShotTraining Approach=Training-free, Shots=0, Backbone=PENGI2026.02 | 0.4965 | |
| Wav2CLIPZero-shot=true2023.05 | 0.414 |