Text-to-audio retrieval on CLOTHO v1 (test)
98SRForget+Forgive
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| Forget+ForgiveTop@K=20, Backbone=WavCaps2026.06 | 98 | 91 | 95 | |
| Forget+ForgiveTop@K=20, Backbone=ATNLL2026.06 | 97 | 80 | 94 | |
| Forget+ForgiveTop@K=10, Backbone=WavCaps2026.06 | 95 | 88 | 91 | |
| Forget+ForgiveTop@K=15, Backbone=ATNLL2026.06 | 95 | 84 | 92 | |
| Forget+ForgiveTop@K=15, Backbone=WavCaps2026.06 | 95 | 88 | 92 | |
| Forget+ForgiveTop@K=20, Backbone=TUAR2026.06 | 94 | 79 | 88 | |
| Forget+ForgiveTop@K=10, Backbone=ATNLL2026.06 | 93 | 75 | 91 | |
| Forget+ForgiveTop@K=15, Backbone=TUAR2026.06 | 92 | 80 | 84 | |
| Forget+ForgiveTop@K=5, Backbone=ATNLL2026.06 | 91 | 60 | 82 | |
| Forget+ForgiveTop@K=5, Backbone=WavCaps2026.06 | 90 | 83 | 86 | |
| Forget+ForgiveTop@K=10, Backbone=TUAR2026.06 | 90 | 78 | 83 | |
| Forget+ForgiveTop@K=5, Backbone=TUAR2026.06 | 89 | 72 | 80 | |
| WavCapsTop@K=20, Debiasing Setting=Baseline2026.06 | 72 | 71 | 74 | |
| ATNLLTop@K=20, Debiasing Setting=Baseline2026.06 | 68 | 52 | 58 | |
| WavCapsTop@K=15, Debiasing Setting=Baseline2026.06 | 67 | 69 | 70 | |
| WavCapsTop@K=10, Debiasing Setting=Baseline2026.06 | 64 | 68 | 69 | |
| ATNLLTop@K=15, Debiasing Setting=Baseline2026.06 | 62 | 50 | 54 | |
| ATNLLTop@K=10, Debiasing Setting=Baseline2026.06 | 60 | 49 | 51 | |
| WavCapsTop@K=5, Debiasing Setting=Baseline2026.06 | 57 | 62 | 52 | |
| TUARTop@K=20, Debiasing Setting=Baseline2026.06 | 54 | 60 | 50 | |
| TUARTop@K=15, Debiasing Setting=Baseline2026.06 | 46 | 58 | 47 | |
| TUARTop@K=10, Debiasing Setting=Baseline2026.06 | 45 | 55 | 46 | |
| TUARTop@K=5, Debiasing Setting=Baseline2026.06 | 41 | 52 | 41 | |
| ATNLLTop@K=5, Debiasing Setting=Baseline2026.06 | 38 | 44 | 46 |