Speech Resynthesis on LibriTTS (test-clean)
0WERReference
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| ReferenceFrame rate=—2026.06 | 0 | — | — | 4.09 | — | 100 | — | — | |
| HybridCodecFrame rate=50 Hz2026.06 | 1.47 | — | — | 4.07 | — | 97.2 | 99.9 | 96.3 | |
| HybridCodecFrame rate=12.5 Hz2026.06 | 1.47 | — | — | 4.09 | — | 96.2 | 97.1 | 96.7 | |
| HybridCodecFrame rate=25 Hz2026.06 | 1.48 | — | — | 4.07 | — | 96.7 | 98.8 | 96.8 | |
| HybridCodecFrame rate=6.25 Hz2026.06 | 1.5 | — | — | 3.98 | — | 97.1 | 97.4 | 98.2 | |
| BigCodecFrame rate=50 Hz2026.06 | 2.55 | — | — | 4.11 | — | 98.5 | 100 | 98.6 | |
| FocalCodecFrame rate=25 Hz2026.06 | 3.3 | — | — | 4.14 | — | 96.3 | 99.8 | 98.4 | |
| DACBitrate (Hz)=86.12026.01 | 3.32 | 0.99 | 4.46 | 3.92 | 1 | — | — | — | |
| FACodecBitrate (Hz)=802026.01 | 3.49 | 0.95 | 2.91 | 4.04 | 0.97 | — | — | — | |
| SpeechTokenizerBitrate (Hz)=502026.01 | 3.53 | 0.93 | 2.61 | 3.8 | 0.96 | — | — | — | |
| MimiBitrate (Hz)=12.52026.01 | 3.59 | 0.97 | 3.47 | 3.85 | 0.97 | — | — | — | |
| Sylber 2.0Bitrate (Hz)=5.812026.01 | 3.86 | 0.89 | 1.99 | 3.8 | 0.92 | — | — | — | |
| SylberBitrate (Hz)=4.222026.01 | 5.44 | 0.75 | 1.13 | 4.09 | 0.76 | — | — | — | |
| MimiFrame rate=12.5 Hz2026.06 | 5.73 | — | — | 3.29 | — | 96 | 95.6 | 91.8 | |
| FocalCodecFrame rate=12.5 Hz2026.06 | 7.94 | — | — | 4.22 | — | 93.9 | 98.2 | 97.4 | |
| WavTokenizerBitrate (Hz)=402026.01 | 16.78 | 0.87 | 1.79 | 3.51 | 0.88 | — | — | — | |
| DACFrame rate=50 Hz2026.06 | 20.04 | — | — | 1.29 | — | 89.2 | 100 | 91.7 |