Audio generation from Encodec tokens on universal evaluation dataset
3.579PESQPeriodWave-Turbo
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| PeriodWave-TurboBandwidth (kbps)=12.0, Parameters=31.3M, Sampling Steps=42025.12 | 3.579 | 4.544 | 0.776 | — | — | |
| Flow2GANBandwidth (kbps)=12.0, Parameters=78.9M, Sampling Steps=42025.12 | 3.538 | 4.531 | 0.557 | — | — | |
| Flow2GANBandwidth (kbps)=12.0, Parameters=78.9M, Sampling Steps=22025.12 | 3.457 | 4.505 | 0.6 | — | — | |
| Flow2GANBandwidth (kbps)=12.0, Parameters=78.9M, Sampling Steps=12025.12 | 3.389 | 4.489 | 0.632 | — | — | |
| PeriodWave-TurboBandwidth (kbps)=6.0, Parameters=31.3M, Sampling Steps=42025.12 | 3.229 | 4.424 | 0.712 | — | — | |
| RFWaveBandwidth (kbps)=12.0, Parameters=18.1M, Sampling Steps=102025.12 | 3.117 | 4.348 | 2.152 | — | — | |
| Flow2GANBandwidth (kbps)=6.0, Parameters=78.9M, Sampling Steps=42025.12 | 3.089 | 4.351 | 0.678 | — | — | |
| Flow2GANBandwidth (kbps)=6.0, Parameters=78.9M, Sampling Steps=22025.12 | 2.983 | 4.319 | 0.733 | — | — | |
| Flow2GANBandwidth (kbps)=6.0, Parameters=78.9M, Sampling Steps=12025.12 | 2.904 | 4.3 | 0.696 | — | — | |
| EnCodecBandwidth (kbps)=12.0, Parameters=56.0M2025.12 | 2.776 | 4.347 | 1.204 | — | — | |
| RFWaveBandwidth (kbps)=6.0, Parameters=18.1M, Sampling Steps=102025.12 | 2.663 | 4.12 | 2.574 | — | — | |
| Flow2GANBandwidth (kbps)=3.0, Parameters=78.9M, Sampling Steps=42025.12 | 2.55 | 4.091 | 0.804 | — | — | |
| Flow2GANBandwidth (kbps)=3.0, Parameters=78.9M, Sampling Steps=22025.12 | 2.442 | 4.049 | 0.843 | — | — | |
| Flow2GANBandwidth (kbps)=3.0, Parameters=78.9M, Sampling Steps=12025.12 | 2.353 | 4.026 | 0.867 | — | — | |
| EnCodecBandwidth (kbps)=6.0, Parameters=56.0M2025.12 | 2.259 | 4.156 | 1.239 | — | — | |
| PeriodWave-TurboBandwidth (kbps)=3.0, Parameters=31.3M, Sampling Steps=42025.12 | 2.16 | 4.058 | 1.018 | — | — | |
| RFWaveBandwidth (kbps)=3.0, Parameters=18.1M, Sampling Steps=102025.12 | 2.124 | 3.784 | 2.622 | — | — | |
| MBDBandwidth (kbps)=6.0, Parameters=411.0M2025.12 | 2.094 | 3.69 | 6.177 | — | — | |
| MBDBandwidth (kbps)=3.0, Parameters=411.0M2025.12 | 1.958 | 3.523 | 6.133 | — | — | |
| Flow2GANBandwidth (kbps)=1.5, Parameters=78.9M, Sampling Steps=42025.12 | 1.925 | 3.662 | 1.069 | — | — | |
| Flow2GANBandwidth (kbps)=1.5, Parameters=78.9M, Sampling Steps=22025.12 | 1.803 | 3.609 | 1.152 | — | — | |
| EnCodecBandwidth (kbps)=3.0, Parameters=56.0M2025.12 | 1.745 | 3.867 | 1.38 | — | — | |
| Flow2GANBandwidth (kbps)=1.5, Parameters=78.9M, Sampling Steps=12025.12 | 1.739 | 3.582 | 1.21 | — | — | |
| RFWaveBandwidth (kbps)=1.5, Parameters=18.1M, Sampling Steps=102025.12 | 1.6 | 3.272 | 2.986 | — | — | |
| MBDBandwidth (kbps)=1.5, Parameters=411.0M2025.12 | 1.457 | 3.03 | 7.734 | — | — | |
| EnCodecBandwidth (kbps)=1.5, Parameters=56.0M2025.12 | 1.368 | 3.409 | 1.996 | — | — | |
| PeriodWave-TurboBandwidth (kbps)=1.5, Parameters=31.3M, Sampling Steps=42025.12 | 1.26 | 3.308 | 4.055 | — | — |