Stability AI optimized its audio generation model to run on Arm chips

Mar 3, 2025

Stability AI partners with Arm to optimize Stable Audio Open for mobile.

Stability AI optimized its audio generation model to run on Arm chips

: Stability AI collaborated with Arm to optimize its audio generation model, Stable Audio Open, for mobile devices. This AI model can operate offline on devices using Arm chips and generates audio from text descriptions, debuting at the Mobile World Congress. Despite the optimized model not yet being available for download, Stability aims to expand its apps and devices offering. CEO Prem Akkaraju emphasizes its potential for creatives and business use, as Stability seeks to recover from prior leadership challenges.

Stability AI has partnered with Arm, a chipmaker, to enhance its AI model, Stable Audio Open, for mobile devices running on Arm chips. This model is unique because it generates audio, like sound effects, from text descriptions and operates offline, unlike other models that require cloud processing.

At the Mobile World Congress conference in Barcelona, the optimized version is capable of creating audio 30 times faster, taking around eight seconds to generate an 11-second sample. Although not yet available for consumer download, CEO Prem Akkaraju highlights plans to integrate such models into consumer apps and devices moving forward.

The collaboration is part of a broader effort by Stability, known for Stable Diffusion, to bounce back from financial challenges. Previous CEO Emad Mostaque's mismanagement led to significant setbacks, but recent strategic changes, including hiring a new CEO and adding James Cameron to its board, highlight Stability's efforts towards recovery.