List your company

Cross-lingual speech generation on Cross-Lingual benchmark

13.5Score (EN->ZH)CosyVoice2
4.42086.77799.13511.4921Jan 22, 2026Feb 5, 2026Feb 19, 2026Mar 5, 2026Mar 19, 2026Apr 2, 2026Apr 17, 2026

Evaluation results

MethodLinks
CosyVoice22026.0113.548.17.76.4717.111.213.114.95.8624.821.921.5
CosyVoice22026.0413.548.17.76.4717.111.213.114.95.8624.821.921.5
Qwen3-TTS-25Hz-1.7B-BaseTokenizer Frequency=25Hz, Number of parameters=1.7B, Model Version=Base2026.015.663.921.142.913.953.489.297.744.178.126.836.86
Qwen3-Omni-30B-A3BBackbone=30B, MoE Config=A3B2026.045.373.320.992.763.313.348.297.534.245.134.966.23
CosyVoice32026.015.093.051.062.984.24.197.086.83.9314.45.877.92
CosyVoice32026.045.093.051.062.984.24.197.086.83.9314.45.877.92
Qwen3.5-Omni-PlusModel Variation=Plus2026.044.863.550.842.182.182.515.925.122.164.033.725.12
Qwen3-TTS-12Hz-1.7B-BaseTokenizer Frequency=12Hz, Number of parameters=1.7B, Model Version=Base2026.014.773.431.082.773.043.098.47.213.674.825.145.59