Fish Speech v1.2
FreeTrained on 300K hours, supports English, Chinese, Japanese.
Voice CloningMultilingualOpen-Source
Overview
Fish Speech v1.2 was trained on roughly 300K hours of audio and delivers strong multilingual performance across English, Chinese and Japanese. It offers fast inference and reliable zero-shot voice cloning. The project is actively maintained with a friendly toolkit for production use.
What makes Fish Speech v1.2 special?
- Focus: Multilingual voice cloning.
- Availability: Free & Open-Source.
- Use cases: Multilingual content, cloning, and production pipelines.