Zero-shot Text-to-Speech on Seed-TTS en (test)
1.08WERQwen3-TTS
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Qwen3-TTSRTF=1.7542026.05 | 1.08 | 71 | — | |
| X-Voices2RTF=0.0732026.05 | 1.3 | 65 | — | |
| Fish Audio S2RTF=4.8012026.05 | 1.37 | 65 | — | |
| Qwen3-TTSSize=1.7B, Emilia=×2026.05 | 1.434 | 71.2 | 4.178 | |
| OmniVoiceSize=0.6B, Emilia=×2026.05 | 1.475 | 74.1 | 3.896 | |
| LEMAS-TTSRTF=0.1312026.05 | 1.49 | 62 | — | |
| X-Voices1RTF=0.0732026.05 | 1.53 | 65 | — | |
| OmniVoiceRTF=0.1982026.05 | 1.6 | 74 | — | |
| Qwen3-TTSSize=0.6B, Emilia=×2026.05 | 1.711 | 70.6 | 4.155 | |
| IndexTTS2Size=1.5B, Emilia=INCL.2026.05 | 1.735 | 70.7 | 3.651 | |
| GibbsTTSSize=0.4B, Emilia=✓2026.05 | 1.777 | 74.3 | 3.651 | |
| F5-TTSRTF=0.0652026.05 | 1.89 | 66 | — | |
| MOSS-TTSRTF=0.6432026.05 | 1.92 | 69 | — | |
| Ground truthSize=—, Emilia=—2026.05 | 2.02 | 73.4 | 3.527 | |
| F5-TTSSize=0.3B, Emilia=✓2026.05 | 2.02 | 65.4 | 3.762 | |
| CosyVoice 3Size=0.5B, Emilia=×2026.05 | 2.145 | 69.6 | 3.949 | |
| F5-TTSmode=zero-shot2026.06 | 2.15 | 75.5 | 3.68 | |
| FireRedTTS-2Size=1.5B, Emilia=×2026.05 | 2.187 | 66.4 | 3.697 | |
| Spark-TTSSize=0.5B, Emilia=✓2026.05 | 2.221 | 57.2 | 3.947 | |
| Codec reconstructedSize=—, Emilia=—2026.05 | 2.229 | 69.5 | 3.407 | |
| CosyVoice 2Size=0.5B, Emilia=×2026.05 | 2.506 | 65.6 | 4.164 | |
| UniVocalmode=zero-shot2026.06 | 2.69 | 70.3 | 4.21 | |
| CosyVoice 2mode=zero-shot2026.06 | 2.96 | 74.4 | 4.18 | |
| MaskGCTSize=1.5B, Emilia=✓2026.05 | 3.763 | 71.6 | 3.582 | |
| Vevo 1.5mode=zero-shot2026.06 | 12.58 | 71.8 | 3.68 |