Text-to-Speech on CosyVoice zh 3 (test)
3.328UTMOSQwen3-TTS
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Qwen3-TTSSize=1.7B, Emilia=×2026.05 | 3.328 | 2.653 | 75.4 | |
| Qwen3-TTSSize=0.6B, Emilia=×2026.05 | 3.221 | 2.9 | 75.5 | |
| CosyVoice 2Size=0.5B, Emilia=×2026.05 | 3.042 | 3.939 | 74.8 | |
| Spark-TTSSize=0.5B, Emilia=✓2026.05 | 2.977 | 4.802 | 67.1 | |
| CosyVoice 3Size=0.5B, Emilia=×2026.05 | 2.899 | 3.166 | 77.6 | |
| OmniVoiceSize=0.6B, Emilia=×2026.05 | 2.884 | 2.735 | 76.6 | |
| IndexTTS2Size=1.5B, Emilia=INCL.2026.05 | 2.544 | 3.064 | 76.2 | |
| GibbsTTSSize=0.4B, Emilia=✓2026.05 | 2.438 | 4.144 | 78 | |
| MaskGCTSize=1.5B, Emilia=✓2026.05 | 2.357 | 6.647 | 77.3 | |
| F5-TTSSize=0.3B, Emilia=✓2026.05 | 2.337 | 5.867 | 74.1 |