Waveform Generation on MUSDB18 OOD Vocal Samples HQ (test)
0.8623M-STFTBigVGAN-v2
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| BigVGAN-v2trained_with_large_scale_dataset=true2024.08 | 0.8623 | 3.47 | 0.084 | 96.49 | 33.229 | |
| BigVSAN2024.08 | 0.8766 | 3.011 | 0.0892 | 95.84 | 37.212 | |
| PeriodWave-Turbo-LSampling steps=4, Sampling algorithm=Euler, Model variant=L2024.08 | 0.8901 | 3.36 | 0.0817 | 95.97 | 28.294 | |
| BigVGAN2024.08 | 0.9062 | 2.862 | 0.0959 | 95.01 | — | |
| BigVGAN2024.08 | 0.9062 | 2.862 | 0.0959 | 95.01 | 30.014 | |
| PeriodWave-Turbo-BSampling steps=4, Sampling algorithm=Euler, Model variant=B2024.08 | 0.9426 | 3.306 | 0.0878 | 95.19 | 31.228 | |
| PeriodWave-Turbo-SSampling steps=4, Sampling algorithm=Euler, Model variant=S2024.08 | 0.983 | 2.98 | 0.0904 | 95.43 | 27.164 | |
| BigVGAN-base2024.08 | 1.0132 | 2.315 | 0.1272 | 93.07 | — | |
| Vocos2024.08 | 1.0203 | 2.173 | 0.1305 | 94.54 | — | |
| Vocos2024.08 | 1.0203 | 2.173 | 0.1305 | 94.54 | 45.631 | |
| PeriodWave-MBSteps=162024.08 | 1.049 | 3.12 | 0.0945 | 95.24 | — | |
| PeriodWave-LSampling steps=16, Sampling algorithm=Midpoint, FreeU enhancement=true2024.08 | 1.0986 | 3.188 | 0.1149 | 94.61 | 36.35 | |
| UnivNet-c322024.08 | 1.1377 | 1.678 | 0.1588 | 91.86 | — | |
| UnivNet-c322024.08 | 1.1377 | 1.678 | 0.1588 | 91.86 | 85.317 | |
| PeriodWaveSteps=16, Sampling strategy=Midpoint, Enhancement=FreeU2024.08 | 1.1923 | 3.062 | 0.0994 | 94.79 | — | |
| PeriodWaveSampling steps=16, Sampling algorithm=Midpoint, FreeU enhancement=true2024.08 | 1.1923 | 3.062 | 0.0994 | 94.79 | 36.715 | |
| PeriodWave-LSampling steps=16, Sampling algorithm=Midpoint2024.08 | 1.1938 | 3.142 | 0.1073 | 93.82 | 29.346 | |
| PeriodWaveSteps=16, Sampling strategy=Midpoint2024.08 | 1.2702 | 2.959 | 0.1046 | 94.75 | — | |
| PeriodWaveSampling steps=16, Sampling algorithm=Midpoint2024.08 | 1.2702 | 2.959 | 0.1046 | 94.75 | 27.012 |