Video-and-Text-to-Audio Generation on Kling-Audio (eval)
2.1KL DivergenceOmni2Sound
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Omni2Sound2026.01 | 2.1 | 7.6 | 2.37 | 106.55 | 5.98 | 8.22 | 58 | 32 | 26 | |
| HunyuanVideo-Foley2026.01 | 2.13 | 8.06 | 3.58 | 94.64 | 6.04 | 8.17 | 55 | 34 | 23 | |
| AudioX2026.01 | 2.39 | 14.26 | 3.16 | 149.37 | 5.97 | 10.23 | 121 | 23 | 26 | |
| MMAudio2026.01 | 2.41 | 10.12 | 4.9 | 129.21 | 5.53 | 7.46 | 59 | 25 | 20 | |
| ThinkSoundCoT (Chain of Thought)=false2026.01 | 2.53 | 11.99 | 3.52 | 206.93 | 5.77 | 6.09 | 66 | 22 | 19 |