Stable Audio Open
audio
open-source
Stable Audio Open (2023) is an open text-to-audio diffusion model by Stability AI. It produces high-quality stereo audio...
Version: 1.0
Released: 2y 1m 18d ago on 09/14/2023
Architecture
- parameters: 0
- context_length: 0
- training_data: Trained on ~48K Creative Commons audio clips (Freesound, Free Music Archive)
- inference: audio diffusion
Capabilities
- Text-to-audio: generates stereo sound or music (up to 47s at 44.1kHz) from text prompts.
- Achieves competitive realism with state-of-the-art audio models.
Benchmarks
- FD_openl3: competitive (state-of-art)
Safety
- Trained on open CC data
- open model may output copyrighted-like audio.
Deployment
- regions: private
- hosting: Hugging Face, API
- integrations: weights downloadable for local use
Tags
audio-generationdiffusionopen-source