Text-to-Speech on Seed-TTS en 24 kHz (test)
0.734SIM-oGround Truth
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Ground Truth2025.12 | 0.734 | 2.14 | 3.52 | |
| E2-TTSNFE=32, Data=100K Emilia, Params=333M2025.12 | 0.706 | 2.32 | 3.21 | |
| ZipVoiceNFE=16, Data=100K Emilia, Params=123M2025.12 | 0.697 | 1.7 | 3.82 | |
| M3-TTS-FbankNFE=32, Data=100K Emilia, Params=355M2025.12 | 0.681 | 1.48 | 3.88 | |
| F5-TTSNFE=32, Data=100K Emilia, Params=336M2025.12 | 0.664 | 1.85 | 3.72 | |
| CosyVoice 2Data=167K Multi., Params=618M2025.12 | 0.652 | 2.57 | — | |
| VAE Reconstruction2025.12 | 0.631 | 1.85 | 2.34 | |
| F5-TTSNFE=32, Data=100K Emilia, Params=155M2025.12 | 0.628 | 1.96 | 3.66 | |
| CosyVoiceData=170K Multi., Params=416M2025.12 | 0.609 | 4.29 | — | |
| M3-TTS-VAENFE=32, Data=100K Emilia, Params=355M2025.12 | 0.604 | 1.36 | 2.8 | |
| Spark-TTSData=102K Multi., Params=507M2025.12 | 0.584 | 1.98 | — |