Stability AI released Stable Audio Open.

Stable Audio Open is a latent diffusion model that generates up to 47 seconds of stereo audio, sound effects, and short musical stems from text prompts.

Hosted on Hugging Face under stabilityai/stable-audio-open-1.0 and on GitHub under Stability-AI/stable-audio-tools.

Released in June 2024.

It gives sound designers and researchers an openly licensed diffusion model trained partly on royalty-free audio for text-to-sound generation.

Implemented in PyTorch via the stable-audio-tools library.

Released under the Stability AI Community License, free for non-commercial and limited commercial use under revenue thresholds; weights gated behind a Hugging Face license click-through.