Speech Reconstruction on LibriTTS 16 kHz (test)
3.56ViSQOLFMelCodec
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| FMelCodecBitrate=250 bps, Sampling Rate=16 kHz2026.05 | 3.56 | 3.48 | 0.92 | 27.01 | 3.6 | 3.72 | 3.51 | |
| MDCTCodecBitrate=250 bps, Sampling Rate=16 kHz2026.05 | 3.45 | 2.43 | 0.92 | 29.27 | 3.32 | 3.13 | 3.11 | |
| BigCodecBitrate=250 bps, Sampling Rate=16 kHz2026.05 | 3.22 | 3.26 | 0.9 | 41.26 | 3.73 | 3.74 | 3.17 | |
| FocalCodecBitrate=250 bps, Sampling Rate=16 kHz2026.05 | 3.12 | 3.26 | 0.92 | 4.97 | 4.35 | 3.65 | 3.31 | |
| DACBitrate=250 bps, Sampling Rate=16 kHz2026.05 | 2.79 | 1.96 | 0.86 | 72.58 | 4.66 | 2.37 | 2.26 | |
| WavTokenizerBitrate=250 bps, Sampling Rate=16 kHz2026.05 | 2.61 | 1.94 | 0.84 | 73.97 | 4.74 | 2.76 | 2.55 | |
| FlowDecBitrate=250 bps, Sampling Rate=16 kHz2026.05 | 2.38 | 1.32 | 0.84 | 76.36 | 5.28 | 1.97 | 2.03 |