Text-to-Speech on Seed ZH-Hard
5.71CERCosyVoice3.5
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CosyVoice3.5Model Architecture Type=AR/Hybrid2026.03 | 5.71 | 78.6 | |
| CosyVoice3-1.5BModel Architecture Type=AR/Hybrid, Number of Parameters=1.5B2026.03 | 5.83 | 75.8 | |
| LongCat-AudioDiT-3.5BNumber of Parameters=3.5B2026.03 | 6.04 | 79.7 | |
| LongCat-AudioDiT-1BNumber of Parameters=1B2026.03 | 6.33 | 78.7 | |
| Qwen3-TTSModel Architecture Type=AR/Hybrid2026.03 | 6.76 | 74.8 | |
| CosyVoice2Model Architecture Type=AR/Hybrid2026.03 | 6.83 | 72.4 | |
| IndexTTS2Model Architecture Type=AR/Hybrid2026.03 | 7.12 | 75.5 | |
| Seed-ICLModel Architecture Type=AR/Hybrid2026.03 | 7.59 | 77.6 | |
| FireRedTTS-1SModel Architecture Type=AR/Hybrid2026.03 | 7.63 | 74.8 | |
| Qwen2.5-OmniModel Architecture Type=AR/Hybrid2026.03 | 7.97 | 74.7 | |
| F5 TTSModel Architecture Type=NAR2026.03 | 8.67 | 71.3 | |
| F5R-TTSModel Architecture Type=NAR2026.03 | 8.79 | 71.8 | |
| VoxCPMModel Architecture Type=AR/Hybrid2026.03 | 8.87 | 73 | |
| MaskGCTModel Architecture Type=NAR2026.03 | 10.27 | 74.8 | |
| CosyVoiceModel Architecture Type=AR/Hybrid2026.03 | 11.75 | 70.9 |