An open-source latent diffusion model that synthesizes complete songs with vocals and instrumentation in under ten seconds.
Audio & Voice

Turn lyrics and musical direction into a complete song with an open full-song generation model.
Best for: Developers, musicians, and technical creators experimenting with local lyrics-to-song generation.
An open-source latent diffusion model that synthesizes complete songs with vocals and instrumentation in under ten seconds.
An open-source generative audio model designed for high-fidelity musical composition and rapid soundscape arrangement.
It writes full songs from a text prompt, including lyrics and vocals, with a license for commercial use.