Riffusion
Riffusion takes a highly unique approach to AI music generation by utilizing Stable Diffusion to generate visual spectrograms, which are then converted back into creative, boundary-pushing audio.
Spectrogram-Based Generation
Unlike traditional audio models, Riffusion generates images (spectrograms) of sound based on your text prompts, resulting in highly experimental and unique musical textures.
Try AI Music Generation
Real-Time Audio Inpainting
Because it operates on images, Riffusion allows you to use standard image inpainting techniques to visually edit the spectrogram, smoothly transitioning between entirely different genres.
Try AI Music GenerationOpen-Source Experimentation
Built on open-source technology, Riffusion invites developers and musicians to experiment with its code, leading to bizarre, wonderful, and highly creative musical outputs.
Try AI Music Generation
How to Use Riffusion?
Enter a Text Prompt
Type a description of the sound or music you want to hear (e.g., "jazzy saxophone over a hip hop beat").
Generate Spectrograms
Riffusion generates an image representation of the sound, which it simultaneously plays back as audio.
Morph and Transition
Input a second prompt and allow the engine to visually and audibly transition smoothly between the two concepts.