Skip to Content

Stable Audio Open Small: Bringing Fast, On-Device AI Audio Generation to Smartphones

Professional Grade Audio Creation on Mobile Devices

Get All The Latest Research & News!

Thanks for registering!

Producing professional-quality sound effects or music loops right from your smartphone—no cloud connection or bulky hardware needed is now possible with the release of Stable Audio Open Small. Developed by Stability AI in partnership with Arm, this compact text-to-audio model democratizes high-quality audio generation, empowering developers and creators to innovate at the edge.

What Sets Stable Audio Open Small Apart?

  • Ultra-Efficient Design: Boasting just 341 million parameters, this model is vastly more lightweight than its predecessor, making it ideal for deployment on smartphones and edge devices.
  • Real-Time Performance: Leveraging Arm’s KleidiAI libraries, it can generate up to 11 seconds of stereo audio in under 8 seconds on a standard mobile device—fast enough for seamless user experiences in apps and games.
  • Maintained Audio Quality: Despite the smaller footprint, output remains crisp and closely aligned with text prompts, supporting a range of creative tasks from drum loops to ambient textures.
  • Truly Open Access: Distributed under Stability AI’s permissive community license, the model is free for both commercial and non-commercial use. Full code, weights, and research documentation are easily accessible on Hugging Face, GitHub, and arXiv.

Strategic Industry Collaboration

This innovation springs from Stability AI’s close work with Arm, whose CPUs power the majority of global smartphones. By optimizing Stable Audio Open Small for these chips, the partnership enables genuine on-device, real-time audio generation—a capability already demonstrated at major events like Mobile World Congress and now available for developers worldwide.

Technical Highlights

  • Mobile-First Architecture: At 341M parameters, Stable Audio Open Small is specifically built for mobile and edge use, in contrast to the much larger 1.1B parameter Stable Audio Open model.
  • Edge Computing Ready: It operates fully on-device, eliminating the need for GPUs or specialized hardware. This reduces latency and costs, making AI-powered audio design accessible and scalable.
  • Optimized for Short Audio Tasks: The model excels at generating concise samples, matching computational resources with real-world creative demands and allowing organizations to choose the right model size for their needs.

Resources for Developers

For those eager to get hands-on, the Arm Learning Path provides step-by-step instructions for practical deployment. Developers seeking deeper technical insights can explore the Arm Community Blog and the official documentation for optimization tips and performance benchmarks.

Enabling the Next Generation of Creative Apps

Stable Audio Open Small is more than a technical milestone—it marks a shift toward real-time, mobile-first creative tools. By lowering the technical and financial barriers to entry, Stability AI and Arm are enabling new possibilities in gaming, media production, and content creation. Users and developers alike benefit from advanced audio synthesis that is accessible, responsive, and scalable.

Takeaway

Stable Audio Open Small stands out as a fast, efficient, and freely available text-to-audio solution for on-device use. Its release is set to inspire a wave of innovation, empowering creators and developers to bring next-generation sound experiences to mobile platforms and beyond.

Source: Stability AI Blog


Stable Audio Open Small: Bringing Fast, On-Device AI Audio Generation to Smartphones
Joshua Berkowitz May 19, 2025
Share this post
Sign in to leave a comment
Cursor’s Fusion Tab Model: AI Code Editing Reimagined
Revolutionizing Code Editing with Predictive Intelligence