Stability AI unveils Stable Audio Open Small, an efficient AI audio generator optimized for smartphones

AI startup Stability AI has introduced Stable Audio Open Small, a stereo audio-generating AI model developed in partnership with Arm, the chipmaker behind many mobile processors. This model is designed to be the fastest on the market and efficient enough to run directly on smartphones without relying on cloud processing, unlike many existing AI audio apps. Stable Audio Open Small is trained exclusively on royalty-free audio from Free Music Archive and Freesound, avoiding copyright issues present in other models. The model consists of 341 million parameters and is optimized for Arm CPUs, capable of generating up to 11 seconds of audio in less than 8 seconds on a smartphone. It is intended for creating short audio clips and sound effects, such as drum and instrument riffs. However, it only supports English prompts and cannot produce realistic vocals or high-quality songs. Its performance varies across musical styles due to Western-biased training data. Usage terms allow free access for researchers, hobbyists, and businesses with under $1 million in revenue, while larger enterprises must pay for a license. Stability AI, known for the Stable Diffusion image model, has faced financial and management challenges recently but is making a comeback with new leadership and product releases.