Open-formed Audio Question Answering on AIR-Bench Sound
7.01ScoreMTC (AudioSet)
Evaluation Results
| Method | Links | |
|---|---|---|
| MTC (AudioSet)Training Data=AudioSet2026.03 | 7.01 | |
| MTC (Ours-UTS)Training Data=UTS2026.03 | 6.68 | |
| Multi-Task (Ours)Training Data=Caption*, Tag*2026.03 | 6.6 | |
| PAR (Ours-UTS)Training Data=UTS2026.03 | 6.59 | |
| Captioning-scratch (Ours)Training Data=Caption*2026.03 | 6.35 | |
| Captioning-scratch [78]Training Data=Original Captions2026.03 | 6.25 | |
| Contrastive-scratch (Ours)Training Data=Caption*2026.03 | 5.78 | |
| Contrastive-scratch [78]Training Data=Original Captions2026.03 | 5.69 |