Speech Reconstruction on VCTK 48 kHz (test)
3.62ViSQOLFMelCodec
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| FMelCodecBitrate=750 bps, Sampling Rate=48 kHz2026.05 | 3.62 | 3.66 | 93 | 4.8 | 2.52 | 3.73 | 3.76 | |
| MDCTCodecBitrate=750 bps, Sampling Rate=48 kHz2026.05 | 3.48 | 3.33 | 92 | 9.62 | 3.13 | 3.48 | 3.73 | |
| BigCodecBitrate=750 bps, Sampling Rate=48 kHz2026.05 | 3.34 | 3.67 | 91 | 10.84 | 2.87 | 3.75 | 3.72 | |
| WavTokenizerBitrate=750 bps, Sampling Rate=48 kHz2026.05 | 3.32 | 2.34 | 75 | 81.76 | 4.41 | 3.52 | 3.3 | |
| DACBitrate=750 bps, Sampling Rate=48 kHz2026.05 | 3.28 | 2.98 | 88 | 33.07 | 3.45 | 3.23 | 3.24 | |
| FlowDecBitrate=750 bps, Sampling Rate=48 kHz2026.05 | 2.77 | 2.93 | 87 | 34.42 | 3.82 | 3.3 | 3.17 |