Text-to-Audio Retrieval on MusicBench (test)
34.52R@1FIGMA
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| FIGMA2026.06 | 34.52 | 65.99 | 81.73 | 91.37 | |
| LAION-CLAPTraining Data=Music, Evaluation Protocol=Zero-shot2026.06 | 25.38 | 55.84 | 68.53 | 79.7 | |
| LAION-CLAPTraining Data=General Audio, Evaluation Protocol=Zero-shot2026.06 | 23.86 | 48.73 | 62.94 | 77.16 | |
| LAION-CLAPTraining Data=General audio with variable-length, Evaluation Protocol=Zero-shot2026.06 | 19.8 | 44.67 | 57.87 | 73.1 | |
| LAION-CLAPTraining Data=Music, Speech and General Audio, Evaluation Protocol=Zero-shot2026.06 | 19.8 | 54.82 | 65.99 | 77.16 | |
| LAION-CLAPTraining Data=Music and Speech, Evaluation Protocol=Zero-shot2026.06 | 19.29 | 46.19 | 64.47 | 77.66 | |
| LAION-CLAPNote=Continued Training on FGMCaps2026.06 | 10.66 | 36.55 | 48.73 | 68.53 |