Stability AI Optimizes Audio Generation for ARM Chips

Stability AI and Arm Collaborate on Mobile Audio Generation

Stability AI, a leading artificial intelligence startup, has forged a partnership with Arm, the prominent chipmaker. This collaboration aims to integrate Stability’s Stable Audio Open, an AI model capable of generating audio – including diverse sound effects – directly onto mobile devices powered by Arm chips.

Addressing Limitations of Existing AI Audio Tools

Numerous AI-driven applications currently offer audio generation capabilities, such as Suno and Udio. However, the majority of these solutions depend on cloud-based processing. This reliance restricts their functionality when an internet connection is unavailable.

Furthermore, concerns exist regarding the training data used by some audio generation models. Certain models have been trained using copyrighted material, potentially creating intellectual property risks.

Stability asserts that the training dataset for Stable Audio Open consists exclusively of royalty-free audio and musical compositions.

Optimized Performance on Arm Architecture

A demonstration of Stable Audio Open running on Arm chips will be showcased at the Mobile World Congress conference in Barcelona this week. The model can produce audio based on textual prompts, for example, “Gentle ocean waves at sunset.”

Stability reports that it collaborated closely with Arm to optimize and “distill” Stable Audio Open. This process has resulted in a 30x acceleration in audio generation speeds.

Specifically, generating a single 11-second audio sample now takes approximately 8 seconds on an Armv9 CPU.

Future Availability and Expansion Plans

Currently, the optimized Stable Audio Open model is not available for direct download. However, Stability CEO Prem Akkaraju indicated that the company intends to make its models, including Stable Audio Open, accessible within consumer applications and on various devices in the future.

“As generative AI becomes increasingly adopted by professional creatives and businesses, ensuring our models and workflows are universally available is crucial for both developers and creators,” Akkaraju stated. “Our partnership with Arm directly addresses this need.”

The companies are continuing to collaborate on further optimization and refinement of Stable Audio Open specifically for mobile platforms.

Stability AI’s Recent Developments

Stability, the company behind the widely-used image generation model Stable Diffusion, secured new funding last year. Investors, including Eric Schmidt and Sean Parker, the founder of Napster, participated in an effort to revitalize the business.

Previous leadership, including co-founder and former CEO Emad Mostaque, reportedly led to financial difficulties for Stability. This resulted in staff departures, a cancelled partnership with Canva, and investor concerns.

In recent months, Stability has appointed a new CEO and added James Cameron, the director of Titanic, to its board of directors. The company has also launched several new image generation models.