Stability AI has introduced a new AI model called Stable Audio Open, designed for generating sounds and music. This model has been trained on royalty-free recordings from Free Sound and the Free Music Archive, and is capable of producing up to 47 seconds of audio based on text descriptions.
According to Stability AI, Stable Audio Open is an open-source model optimized for generating short audio samples, sound effects, and production elements using text prompts. This release is a significant milestone in opening up generative audio capabilities to empower sound designers, musicians, and creative communities.
With a training set of around 486,000 samples of royalty-free music and sound libraries, Stable Audio Open aims to provide a versatile tool for creating various audio elements based on a text input. Users can generate instrumentals, drum beats, ambient noise, and other audio production elements for use in videos, film, and television.
The tool is intended to be a valuable resource for sound designers, musicians, and creative professionals, allowing them to create high-quality audio from a simple text prompt. The royalty-free training of Stable Audio Open makes it particularly useful for creating sounds for music production and sound design.
Stability AI encourages users to download the Stable Audio Open model, explore its capabilities, and provide feedback. The company sees this as just the beginning for open and responsible audio generation capabilities, and looks forward to further research and development in collaboration with creative communities.
Stability AI also offers a commercial model of Stable Audio for producing full tracks with coherent musical structure up to three minutes in length. Stable Audio Open, on the other hand, is not optimized for full songs, melodies, or vocals. The company emphasizes responsible development and sees the open model as a glimpse into generative AI for sound design.
