Text-to-Audio Retrieval on Internal professional sound effects dataset ID (eval)
75.4R@10Cards-CLAP
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Cards-CLAPTraining regime=Audiocard fields, Evaluation mode=Zero-shot2026.02 | 75.4 | 78.73 | |
| Captions-CLAPTraining regime=Baseline captions, Evaluation mode=Zero-shot2026.02 | 73.45 | 77.66 | |
| LAION-CLAPTraining regime=None specified, Evaluation mode=Zero-shot2026.02 | 24.85 | 47.1 |