| Dataset Name | SOTA Method | Metric | Trend | ||
|---|---|---|---|---|---|
| LibriTTS (CLEAN), LibriVox (NOISY), YouTube (WILD), and My Science Tutor (KIDS) (test) | MOS3.7 | 21 | 3mo ago | ||
| LJSpeech | FNH-TTS | MOS4.48 | 11 | 5d ago | |
| VCTK | FNH-TTS | MOS4.63 | 9 | 5d ago | |
| 10-second speech segments | MELLE-R4 | Inference Time (s)1.4 | 8 | 3mo ago | |
| SEED-TTS en | zh | avg. | UniAudio-Token | SIM (en)0.792 | 2 | 2d ago |