DiffRythm is an AI model for generating music, developed by a Chinese research team. It can create multi-minute tracks in seconds while keeping audio clear and consistent.
What DiffRythm does
DiffRythm focuses on fast music generation and supports timing adjustments, which can help align lines or sections more precisely.
Generates tracks up to about 4 minutes 45 seconds in ~10 seconds
Clear audio with minimal distortion, even on longer outputs
Timing (line/section) editing support
Style control through text prompts
Free access
How to use DiffRythm
DiffRythm runs as a web app on Hugging Face, so you don’t need to install anything. It can handle prompts in Russian as well as English and other languages.
Open the official Hugging Face page
Choose settings or upload inputs (if available)
Start generation
Wait for processing to finish and download the result
Technical notes
Uses latent diffusion for music synthesis
Open model with source code available
Works in the browser via a web interface
