Stability AI Launches Smartphone-Compatible Audio Generation Model

AI startup Stability AI has launched Stable Audio Open Small, a stereo audio generation model that it claims is the fastest available and efficient enough to operate on smartphones.

Stable Audio Open Small results from a partnership between Stability AI and Arm, the chipmaker behind many processors in tablets, phones, and other mobile devices. While numerous AI apps like Suno and Udio can generate audio, most depend on cloud processing and cannot function offline.

Stability also states that Stable Audio Open Small's training dataset consists entirely of royalty-free tracks from the Free Music Archive and Freesound libraries. This contrasts with the training sets of Suno and Udio, which are said to include copyrighted material, creating potential intellectual property risks.

With 341 million parameters and optimization for Arm CPUs, Stable Audio Open Small is designed for rapid creation of short audio clips and sound effects such as drum patterns and instrument riffs. Stability AI claims it can generate up to 11 seconds of audio on a smartphone in under 8 seconds.

Listen to this sample created with Stable Audio Open Small:

Here is another example:

The model does have limitations. Stable Audio Open Small only accepts prompts in English, and Stability notes in its documentation that it cannot produce realistic vocals or high-fidelity songs. Performance also varies across musical genres due to its Western-centric training data.

Join TechCrunch Sessions: AI

Secure your place at our premier AI industry event featuring speakers from OpenAI, Anthropic, and Cohere. For a limited time, attend a full day of expert talks, workshops, and valuable networking for just $292.

Exhibit at TechCrunch Sessions: AI

Reserve your exhibit space at TC Sessions: AI and present your innovations to over 1,200 decision-makers without a large investment. Available until May 9 or while supplies last.

Developers should also note that Stable Audio Open Small comes with somewhat restrictive usage terms. It is free for researchers, hobbyists, and businesses with under $1 million in annual revenue, but developers and organizations exceeding that revenue must purchase Stability's enterprise license.

Stability, the company behind the widely used image generation model Stable Diffusion, secured additional funding last year from investors including Eric Schmidt and Napster founder Sean Parker to help stabilize the business. Former CEO and co-founder Emad Mostaque's alleged mismanagement reportedly led to financial difficulties, staff departures, a failed partnership with Canva, and investor concerns about the company's future.

Recently, Stability has appointed a new CEO, added Titanic director James Cameron to its board, and released multiple new image generation models.