Text-to-Audio Retrieval on MECAT (test)
8.02Recall@1MGA-CLAP
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| MGA-CLAPModel Type=CLAP Models2026.04 | 8.02 | 23.51 | 34.08 | |
| OEA-Nemo3BModel Type=OEA Models, Backbone=Nemotron-3B2026.04 | 7.96 | 24.53 | 35.81 | |
| OEA-Qwen7BModel Type=OEA Models, Backbone=Qwen7B2026.04 | 7.74 | 23.29 | 33.49 | |
| M2D-CLAPModel Type=CLAP Models2026.04 | 7.37 | 23.55 | 33.63 | |
| OEA-Qwen7B (+Cl)Model Type=OEA Models, Backbone=Qwen7B, Additional Training=Clotho2026.04 | 7.02 | 22.94 | 33.12 | |
| OEA-Nemo3B (+Cl)Model Type=OEA Models, Backbone=Nemotron-3B, Additional Training=Clotho2026.04 | 6.98 | 23 | 33.41 | |
| LAION-CLAPModel Type=CLAP Models2026.04 | 6.8 | 20.93 | 30.66 | |
| Robust-CLAPModel Type=CLAP Models2026.04 | 6.51 | 21.05 | 31.39 | |
| OEA-Qwen3BModel Type=OEA Models, Backbone=Qwen3B2026.04 | 5.86 | 17.96 | 25.94 | |
| OEA-Qwen3B (+Cl)Model Type=OEA Models, Backbone=Qwen3B, Additional Training=Clotho2026.04 | 5.78 | 17.16 | 24.76 | |
| Nemotron-3BModel Type=Vanilla LALMs, Retrieval Training=No2026.04 | 4.09 | 13.27 | 21.99 | |
| Qwen2.5-Omni-7BModel Type=Vanilla LALMs, Retrieval Training=No2026.04 | 0.16 | 0.77 | 2.38 | |
| Qwen2.5-Omni-3BModel Type=Vanilla LALMs, Retrieval Training=No2026.04 | 0.12 | 0.63 | 1.57 |