Speech Synthesis on ESD Zh
2.4WERSpark-TTS
Evaluation results
| Method | Links | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Spark-TTSBackbone=Qwen2.52025.10 | 2.4 | 0.7 | 6.2 | 5.99 | 82.29 | 5.56 | 0 | 4 | 0 | 18.37 | |
| RLAIF-SPABackbone=MiniCPM-O 2.62025.10 | 3.68 | 0.74 | 6.5 | 6.29 | 89.71 | 7.14 | 4.57 | 4.86 | 3.14 | 21.88 | |
| MegaTTS32025.10 | 3.86 | 0.72 | 6.4 | 6.28 | 54 | 51.14 | 0 | 0.86 | 0 | 21.2 | |
| F5-TTS2025.10 | 4.01 | 0.69 | 6.45 | 6.06 | 87.71 | 0 | 0 | 10.29 | 0.57 | 19.71 | |
| Chat-TTS2025.10 | 6.7 | 0.67 | 6.06 | 5.92 | 63.14 | 6.35 | 2.57 | 2.57 | 2.86 | 15.5 |