Text-to-Speech on LibriSpeech-PC subset B
1.95WERCosyVoice3
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| CosyVoice3Type=AR, Params.=0.5B, Dataset=1000k Multi.2026.06 | 1.95 | 71.8 | 4.28 | |
| CosyVoice2Type=AR, Params.=0.5B, Dataset=170k Multi.2026.06 | 2.05 | 65.7 | 4.38 | |
| TLDRType=Ours, k=4, Params.=0.5B + 136.2M, Dataset=0.6k LibriTTS2026.06 | 2.15 | 71 | 4.24 | |
| TLDRType=Ours, k=6, Params.=0.5B + 136.2M, Dataset=0.6k LibriTTS2026.06 | 2.2 | 70.8 | 4.23 | |
| Ground-truthParams.=-, Dataset=-2026.06 | 2.23 | 69 | 4.1 | |
| DiTARType=AR, Params.=0.6B, Dataset=100k Emilia2026.06 | 2.39 | 67 | 4.22 | |
| F5-TTSType=NAR, NFE=32, Params.=0.3B, Dataset=100k Emilia2026.06 | 2.42 | 66 | 3.88 | |
| TLDRType=Ours, k=8, Params.=0.5B + 136.2M, Dataset=0.6k LibriTTS2026.06 | 2.53 | 70.9 | 4.23 | |
| FireRedTTSType=AR, Params.=0.6B, Dataset=248k Multi.2026.06 | 2.69 | 47 | — | |
| MaskGCTType=NAR, NFE=50, Params.=1.1B, Dataset=100k Emilia2026.06 | 2.72 | 69 | 3.9 | |
| E2 TTSType=NAR, NFE=32, Params.=0.3B, Dataset=100k Emilia2026.06 | 2.95 | 69 | 3.56 | |
| CosyVoiceType=AR, Params.=0.3B, Dataset=170k Multi.2026.06 | 3.59 | 66 | 4.14 |