Speech Synthesis on Seed-TTS-Eval en (test)
1.85WERVoxCPM
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VoxCPMType=AR2026.03 | 1.85 | 72.9 | |
| LLaDA-TTSType=Diffusion, Steps=64 steps2026.03 | 1.96 | 71.1 | |
| F5-TTSType=Flow2026.03 | 2 | 64.7 | |
| HumanType=—2026.03 | 2.14 | 73.4 | |
| Index-TTS 2Type=AR2026.03 | 2.23 | 70.6 | |
| CosyVoice 3-0.5BType=AR2026.03 | 2.24 | 71.8 | |
| Seed-TTSType=AR2026.03 | 2.25 | 76.2 | |
| MaskGCTType=NAR2026.03 | 2.62 | 71.4 |