Text-to-Speech on Seed-TTS-Eval zh (test)
0.93CERVoxCPM
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| VoxCPMType=AR2026.03 | 0.93 | 77.2 | |
| LLaDA-TTSType=Diffusion, Steps=64 steps2026.03 | 0.98 | 74.6 | |
| Index-TTS 2Type=AR2026.03 | 1.03 | 76.5 | |
| Seed-TTSAccess=Proprietary2026.02 | 1.12 | — | |
| Seed-TTSType=AR2026.03 | 1.12 | 79.6 | |
| CosyEdit2mode=zero-shot2026.05 | 1.16 | 75.2 | |
| Spark TTSAccess=Open-Source2026.02 | 1.2 | — | |
| CosyVoice 3-0.5BType=AR2026.03 | 1.21 | 78 | |
| Human2026.02 | 1.25 | — | |
| HumanType=—2026.03 | 1.26 | 75.5 | |
| CosyVoice2mode=zero-shot2026.05 | 1.36 | 75.1 | |
| CosyVoice 2Access=Open-Source2026.02 | 1.45 | — | |
| FireRedTTSAccess=Open-Source2026.02 | 1.51 | — | |
| F5-TTSType=Flow2026.03 | 1.52 | 74.1 | |
| Qwen2.5-OmniAccess=Open-Source2026.02 | 1.7 | — | |
| CosyEditmode=zero-shot2026.05 | 1.76 | 74.7 | |
| MaskGCTType=NAR2026.03 | 2.27 | 77.4 | |
| VoiceCraft-Xmode=zero-shot2026.05 | 3.26 | 67.6 | |
| Ex-OmniAccess=Open-Source2026.02 | 3.37 | — | |
| CosyVoiceAccess=Open-Source2026.02 | 3.63 | — | |
| SSR-Speechmode=zero-shot2026.05 | 20.51 | 53.9 |