Enhanced features of the generative AI “Stable Audio Open,” generating music in under 8 seconds on offline smartphones—how to utilize Arm’s technology? [Japanese]

Author | Source | Read the full article in Japanese

Stability AI has recently announced an exciting enhancement to its music generation AI, known as "Stable Audio Open," in collaboration with Arm. This new feature allows users to create music directly on their mobile devices without needing an internet connection. This means that anyone with a smartphone equipped with an Arm CPU can generate music quickly and efficiently, making the process much more accessible.

The Stable Audio Open model can produce audio samples and sound effects of up to 47 seconds in length, all generated from text input. Users can also customize the model using their own data, thanks to the availability of model weights on Hugging Face. This flexibility allows for a wide range of creative possibilities, enabling users to tailor the music generation to their specific needs.

According to Stability AI, the integration of Arm's "KleidiAI" library has significantly improved the performance of the music generation process. The new setup is reported to be 30 times faster than previous versions, reducing the time it takes to create music from several minutes to just a few seconds. This advancement opens up new opportunities for musicians and creators to experiment with sound in real-time, even when offline.

[Read More (translated)]