Text-and-Video-to-Audio (TV2A) on VGGSound
1.51KL DivergenceMMAudio
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| MMAudioTask=TV2A2026.06 | 1.51 | 17.79 | 6.6 | 2.2 | 3.31 | 5.99 | 33 | |
| AudioX-BaseTask=TV2A2026.06 | 1.56 | 17.81 | 6.59 | 1.86 | 3.66 | 6.34 | 28 | |
| AudioX-TurboTask=TV2A2026.06 | 1.6 | 17.47 | 5.78 | 0.97 | 3.67 | 6.25 | 29 | |
| FoleyCrafterTask=TV2A2026.06 | 1.94 | 11.32 | 19.16 | 2.13 | 3.38 | 6.06 | 26 |