Text-to-Speech on Seed-TTS-Eval (test)
1.33WERHoliTok
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| HoliTokZero-shot=true2026.05 | 1.33 | 0.62 | |
| RobustSpeechFlowParams=0.06B, Zero-shot=true2026.05 | 1.38 | 0.6 | |
| ContrastiveFMParams=0.06B, Zero-shot=true2026.05 | 1.41 | 0.6 | |
| Semantic-VAEZero-shot=true2026.05 | 1.42 | 0.63 | |
| Baseline(SupertonicTTS)Params=0.06B, Zero-shot=true2026.05 | 1.44 | 0.6 | |
| Qwen3-TTSParameters=1.7B, Training Data=~ 5M, Open-Weight=true, Open-Data=false2026.05 | 1.46 | 0.715 | |
| MiniMax-SpeechZero-shot=true2026.05 | 1.65 | 0.69 | |
| DiTARParams=0.6B, Zero-shot=true2026.05 | 1.69 | 0.74 | |
| Seed-TTSDiTZero-shot=true2026.05 | 1.73 | 0.79 | |
| Raon-OpenTTS-1BParameters=1.0B, Training Data=510K, Open-Weight=true, Open-Data=true2026.05 | 1.78 | 0.749 | |
| MingTok-AudioZero-shot=true2026.05 | 1.84 | 0.61 | |
| VoxCPMParams=0.5B, Zero-shot=true2026.05 | 1.85 | 0.73 | |
| OpenAudio S1-MiniParams=0.5B, Zero-shot=true2026.05 | 1.94 | 0.55 | |
| Raon-OpenTTS-0.3BParameters=0.3B, Training Data=510K, Open-Weight=true, Open-Data=true2026.05 | 1.95 | 0.687 | |
| VoxCPMParameters=0.5B, Training Data=1.8M, Open-Weight=true, Open-Data=false2026.05 | 1.98 | 0.73 | |
| F5-TTSParams=0.3B, Zero-shot=true2026.05 | 2 | 0.67 | |
| F5-TTSParameters=0.3B, Training Data=100K, Open-Weight=true, Open-Data=true2026.05 | 2.04 | 0.671 | |
| Human2026.05 | 2.14 | 0.734 | |
| Index-TTS 2Parameters=1.5B, Training Data=55K, Open-Weight=true, Open-Data=false2026.05 | 2.18 | 0.709 | |
| Voxtral TTSParameters=4B, Open-Weight=true, Open-Data=false2026.05 | 2.19 | 0.663 | |
| CosyVoice 3Parameters=1.5B, Training Data=~ 1M, Open-Weight=false, Open-Data=false2026.05 | 2.21 | 0.72 | |
| CosyVoice3Params=1.5B, Zero-shot=true2026.05 | 2.22 | 0.72 | |
| IndexTTS2Params=1.5B, Zero-shot=true2026.05 | 2.23 | 0.71 | |
| Seed-TTS2026.05 | 2.25 | 0.762 | |
| VoxCPM-EmiliaParams=0.5B, Zero-shot=true2026.05 | 2.34 | 0.68 | |
| CosyVoice 3Parameters=0.5B, Training Data=~ 1M, Open-Weight=true, Open-Data=false2026.05 | 2.5 | 0.698 | |
| MaskGCTParameters=0.6B, Training Data=100K, Open-Weight=true, Open-Data=true2026.05 | 2.57 | 0.713 | |
| CosyVoice 2Parameters=0.5B, Training Data=170K, Open-Weight=true, Open-Data=false2026.05 | 2.61 | 0.659 | |
| MegaTTS3Params=0.5B, Zero-shot=true2026.05 | 2.79 | 0.77 | |
| VibeVoiceParams=1.5B, Zero-shot=true2026.05 | 3.04 | 0.69 | |
| Spark-TTSParams=0.5B, Zero-shot=true2026.05 | 3.14 | 0.57 | |
| LlasaParameters=8B, Training Data=250K, Open-Weight=true, Open-Data=false2026.05 | 3.63 | 0.581 | |
| CleanCodecToken Rate (t/s)=12.5, Train Time (hh:mm)=01:00, RTF (RTFx)=0.170 (5.9x)2026.06 | 3.9 | 0.56 | |
| XCodec2Token Rate (t/s)=50, Train Time (hh:mm)=02:44, RTF (RTFx)=0.688 (1.5x)2026.06 | 4.9 | 0.4 | |
| BiCodecToken Rate (t/s)=50, Train Time (hh:mm)=02:49, RTF (RTFx)=0.691 (1.4x)2026.06 | 5.2 | 0.51 | |
| KanadeToken Rate (t/s)=12.5, Train Time (hh:mm)=00:59, RTF (RTFx)=0.169 (5.9x)2026.06 | 5.6 | 0.45 | |
| Qwen3Token Rate (t/s)=200, Train Time (hh:mm)=10:02, RTF (RTFx)=2.930 (0.3x)2026.06 | 9.1 | 0.36 | |
| MimiToken Rate (t/s)=100, Train Time (hh:mm)=05:11, RTF (RTFx)=1.452 (0.7x)2026.06 | 10.8 | 0.34 | |
| FocalCodecToken Rate (t/s)=12.5, Train Time (hh:mm)=00:58, RTF (RTFx)=0.172 (5.8x)2026.06 | 12.2 | 0.31 | |
| WavTokenizerToken Rate (t/s)=40, Train Time (hh:mm)=02:19, RTF (RTFx)=0.552 (1.8x)2026.06 | 16.6 | 0.37 |