Speech Synthesis on MELD En
8.92WERRLAIF-SPA
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| RLAIF-SPABackbone=MiniCPM-O 2.62025.10 | 8.92 | 0.55 | 6.73 | 6.3 | 68.65 | 29.76 | 41.6 | 5.73 | 29.44 | 35.04 | |
| F5-TTS2025.10 | 9.07 | 0.44 | 6.63 | 6.1 | 71.57 | 24.6 | 14.4 | 36.26 | 15.56 | 32.48 | |
| MegaTTS32025.10 | 9.57 | 0.53 | 6.58 | 6.25 | 67.13 | 22.22 | 29.6 | 3.82 | 28.89 | 30.33 | |
| Spark-TTSBackbone=Qwen2.52025.10 | 15.54 | 0.53 | 6.18 | 5.97 | 27.79 | 50.79 | 9.6 | 2.29 | 38.33 | 25.76 | |
| Chat-TTS2025.10 | 19.64 | 0.54 | 6.17 | 5.85 | 61.17 | 29.37 | 21.6 | 3.44 | 20.56 | 27.23 |