Overview
Fish Speech is the open codebase behind Fish Audio’s speech research and product work. Its documentation covers installation, model downloads, command-line inference, an API server, a web interface, and agent-oriented usage. The project targets multilingual speech generation and supports voice cloning workflows using reference audio. The repository and associated model weights use the Fish Audio Community License, while S2-Pro weights have a separate Fish Audio Research License. Because these are custom terms rather than an OSI-approved license, the product is not marked as open source under this directory’s rule, despite its public code and active development.