Custom-voice Speech Generation on Multilingual (test)
0.785Chinese WERQwen3.5-Omni-Plus
Evaluation Results
| Method | Links | |||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Qwen3.5-Omni-PlusCapability=Custom-voice, Setting=Multilingual2026.04 | 0.785 | 0.839 | 0.182 | 0.458 | 1.581 | 0.768 | 3.306 | 1.309 | 2.724 | 4.723 | 1.653 | 1.596 | 3.183 | 1.32 | 1.309 | 4.039 | 1.462 | 6.776 | 1.135 | 3.769 | 14.916 | 5.09 | 3.588 | 7.183 | 7.68 | 10.322 | 3.738 | 5.576 | 12.14 | |
| MiniMaxCapability=Custom-voice, Setting=Multilingual2026.04 | 0.786 | 1.429 | 1.581 | 1.063 | 1.24 | 0.691 | 4.254 | 3.635 | 3.439 | 3.529 | 1.811 | 1.585 | 3.309 | 1.058 | 0.652 | 2.939 | 0.833 | 6.146 | 1.406 | 1.766 | 24.151 | 5.674 | 2.833 | 4.951 | 8.161 | 33.431 | 3.955 | 9.492 | 12.722 | |
| GPT-AudioCapability=Custom-voice, Setting=Multilingual2026.04 | 0.829 | 1.05 | 0.558 | 0.769 | 1.506 | 0.936 | 4.317 | 3.999 | 2.809 | 4.346 | 4.43 | 2.362 | 5.326 | 1.854 | 1.389 | 3.27 | 1.737 | 7.191 | 1.561 | 2.859 | 13.362 | 5.352 | 2.898 | 3.846 | 5.328 | 9.648 | 3.406 | 3.4 | 13.202 | |
| Gemini-2.5 ProCapability=Custom-voice, Setting=Multilingual2026.04 | 1.89 | 0.953 | 0.509 | 0.991 | 2.05 | 0.891 | 4.42 | 4.11 | 3.284 | 3.858 | 2.539 | 1.498 | 5.525 | 1.699 | 2.237 | 5.331 | 1.622 | 6.596 | 0.973 | 3.38 | 14.141 | 6.784 | 3.196 | 3.876 | 4.459 | 6.348 | 3.731 | 4.304 | 12.62 | |
| ElevenLabsCapability=Custom-voice, Setting=Multilingual2026.04 | 3.801 | 1.126 | 0.5 | 0.513 | 1.109 | 0.52 | 11.685 | 3.981 | 2.574 | 4.324 | 114.813 | 6.094 | 5.4 | 82.849 | 0.551 | 2.522 | 0.733 | 6.388 | 1.005 | 1.916 | 12.97 | 5.473 | 3.132 | 2.604 | 102.018 | 25.11 | 6.448 | 7.351 | 20.564 |