Text-to-Speech on Seed-TTS-Eval zh-hard (test)
6.71CERCosyVoice 3-0.5B
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CosyVoice 3-0.5BType=AR2026.03 | 6.71 | 75.8 | |
| LLaDA-TTSType=Diffusion, Steps=64 steps2026.03 | 7.04 | 73.5 | |
| Index-TTS 2Type=AR2026.03 | 7.12 | 75.5 | |
| Seed-TTSType=AR2026.03 | 7.59 | 77.6 | |
| F5-TTSType=Flow2026.03 | 8.67 | 71.3 | |
| VoxCPMType=AR2026.03 | 8.87 | 73 | |
| MaskGCTType=NAR2026.03 | 10.27 | 74.8 |