Ali Tongyi Bailing Unveils New Iteration of Its Voice Model
2025-12-15 / Read about 0 minute
Author:小编   

Ali Tongyi Lab has made a significant announcement, revealing the open-sourcing of its latest innovation, Fun-CosyVoice3-0.5B. This cutting-edge version boasts an impressive zero-shot voice cloning feature. What does this mean for users? Simply put, by providing a reference audio clip that's over 3 seconds long, users can effortlessly replicate a voice and generate entirely new speech. Furthermore, the model is designed with user flexibility and convenience in mind, supporting local deployment and facilitating secondary development for those who wish to customize and enhance their experience.