Dialect Synthesis on Dialect Synthesis Dataset (Guangdong, Shanghai, Sichuan, and Shaanxi dialects) (test)
7.43Avg WERFT-Last
Evaluation results
| Method | Links | ||
|---|---|---|---|
| FT-LastFine-tuning=Last layer2025.12 | 7.43 | 0.65 | |
| FTFine-tuning=Full2025.12 | 9.04 | 0.72 | |
| CosyVoice22025.12 | 14.49 | 0.63 | |
| E-Vector2025.12 | 15.41 | 0.65 | |
| GT2025.12 | 16.59 | — | |
| LoRA E-Vectoradaptation=LoRA2025.12 | 18.58 | 0.7 |