Video-to-audio generation on OGameData (test)
0FDGround truth
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Ground truthtext-conditioned=true2026.04 | 0 | 0 | 0.311 | 0.942 | — | |
| Woosh-DVFlowParams=413M, text-conditioned=true2026.04 | 10.859 | 1.076 | 0.304 | 0.922 | 0.197 | |
| Woosh-VFlowParams=413M, text-conditioned=true2026.04 | 11.15 | 1.006 | 0.327 | 0.913 | 0.202 | |
| Woosh-VFlowParams=413M, text-conditioned=false2026.04 | 11.793 | 1.376 | 0.336 | 0.893 | 0.123 | |
| Woosh-DVFlowParams=413M, text-conditioned=false2026.04 | 12.239 | 1.435 | 0.313 | 0.913 | 0.126 | |
| MMAudio-MParams=621M, text-conditioned=false2026.04 | 86.746 | 2.758 | 0.207 | 0.926 | 0.107 | |
| MMAudio-MParams=621M, text-conditioned=true2026.04 | 87.186 | 2.267 | 0.215 | 0.927 | 0.24 |