I Built a Local AI Music Studio

Developer Sizzlebop has introduced Miso, an open-source, self-hosted local music generation and remix studio. Built on top of audio.cpp, Miso provides a comprehensive interface for managing music models, allowing users to move beyond simple text-to-song generation. The platform supports various tasks, including stem separation, voice conversion, audio-to-MIDI transcription, and track repainting. Miso is designed to be local-first, running via Docker containers and utilizing a React-based frontend with a SQLite backend. By integrating multiple models like ACE-Step 1.5, YuE2, and Stable Audio 3, the project offers a unified workflow for complex audio experimentation. The developer emphasizes that the tool focuses on the creative process after initial generation, such as editing, mixing, and rearranging stems. Miso is currently available on GitHub for users looking to experiment with local AI music production without relying on cloud-based services or command-line interfaces.