Audio-to-Text Retrieval on ESC50
95Recall@1Proposed
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| ProposedModality=a2t, audio encoder=HTSAT-tiny, text encoder=RoBERTa-large2026.04 | 95 | 99.5 | 99.8 | 97.2 | |
| Microsoft-CLAPModality=a2t2026.04 | 93.5 | 99.8 | 100 | 95.6 | |
| LAION-CLAPModality=a2t2026.04 | 91.5 | 99.5 | 99.7 | 94.7 |