Instance Retrieval on UCS corpus (unseen sound events)
98.8R@1M2D
Evaluation Results
| Method | Links | |
|---|---|---|
| M2Devaluation_protocol=instance head, pretraining=self-supervised (AS2M)2026.07 | 98.8 | |
| BEATsevaluation_protocol=instance head, pretraining=self-supervised (AS2M)2026.07 | 96.6 | |
| AudioMAEevaluation_protocol=instance head, pretraining=self-supervised (AS2M)2026.07 | 96.6 | |
| ASTevaluation_protocol=instance head, pretraining=supervised (AudioSet)2026.07 | 94 | |
| M2Devaluation_protocol=frozen, pretraining=self-supervised (AS2M)2026.07 | 89.5 | |
| BEATsevaluation_protocol=frozen, pretraining=self-supervised (AS2M)2026.07 | 82.6 | |
| ASTevaluation_protocol=frozen, pretraining=supervised (AudioSet)2026.07 | 81.5 | |
| AudioMAEevaluation_protocol=frozen, pretraining=self-supervised (AS2M)2026.07 | 80.4 | |
| CLAPevaluation_protocol=instance head, pretraining=audio-text contrastive2026.07 | 80 | |
| PANNs CNN14evaluation_protocol=instance head, pretraining=supervised (AudioSet)2026.07 | 73.1 | |
| PANNs CNN14evaluation_protocol=frozen, pretraining=supervised (AudioSet)2026.07 | 65.7 | |
| CLAPevaluation_protocol=frozen, pretraining=audio-text contrastive2026.07 | 61.1 | |
| M2Devaluation_protocol=class head, pretraining=self-supervised (AS2M)2026.07 | 48.6 | |
| AudioMAEevaluation_protocol=class head, pretraining=self-supervised (AS2M)2026.07 | 45.2 | |
| BEATsevaluation_protocol=class head, pretraining=self-supervised (AS2M)2026.07 | 33.5 | |
| PANNs CNN14evaluation_protocol=class head, pretraining=supervised (AudioSet)2026.07 | 31.6 | |
| ASTevaluation_protocol=class head, pretraining=supervised (AudioSet)2026.07 | 29.5 | |
| CLAPevaluation_protocol=class head, pretraining=audio-text contrastive2026.07 | 26.9 |