Mixed-Audio Generation on MECAT Speech + Audio (S0A)
30.38FADVGGMusicGen
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| MusicGen2026.05 | 30.38 | 89.09 | 5.6 | 0.2 | -13.67 | 100 | 1.65 | |
| MusicGen2026.05 | 20.18 | 69.19 | 3.41 | 10.7 | -8.04 | 100 | 1.69 | |
| Qwen3-TTS2026.05 | 19.62 | 62.84 | 2.42 | 15.3 | -6.51 | 13.63 | 3.48 | |
| Qwen3-TTS2026.05 | 16.53 | 83.01 | 2.45 | 10.2 | -8.05 | 14.92 | 3.22 | |
| TangoFlux2026.05 | 11.4 | 50.58 | 2.05 | 29 | 1.18 | 99.72 | 1.63 | |
| TangoFlux2026.05 | 10.92 | 43.94 | 2.23 | 36 | 3.99 | 99.3 | 1.84 | |
| Expert-Pipeline2026.05 | 7.1 | 32.36 | 1.86 | 37.9 | 6.5 | 49.22 | 2.22 | |
| Expert-Pipeline2026.05 | 6.38 | 30.1 | 1.08 | 37.7 | 7.32 | 62.14 | 2.24 | |
| Dasheng AudioGen2026.05 | 2.17 | 17.75 | 0.63 | 38.3 | 9.52 | 28.98 | 2.46 | |
| Dasheng AudioGen2026.05 | 1.75 | 8.56 | 0.69 | 36.3 | 11.08 | 22.98 | 2.6 |