Deploying GenAI for Audio Generation on Mobile CPUs With ExecuTorch

Deploying GenAI for Audio Generation on Mobile CPUs With ExecuTorch - Gian Marco Iodice, Arm Generative AI (GenAI) is often associated with cloud-based large language models (LLMs), but its capabilities and deployment scenarios extend far beyond this narrow view. Today, GenAI is already finding its place on mobile and edge devices—operating without internet connectivity to enhance messaging apps, productivity tools, and more. This talk expands the common view of GenAI beyond text generation, highlighting its broader potential, including real-time audio and media generation on-device. We will present practical use cases of GenAI running locally on mobile platforms and focus on the technical details of deploying these models efficiently using ExecuTorch on Arm CPUs. Attendees will gain insight into the optimizations, and integration workflows that directly enable responsive, private, and intelligent experiences on edge devices.