Archive — US — techcrunch
Stability AI optimized its audio generation model to run on Arm chips
Stability AI has partnered with Arm to optimize its Stable Audio Open model for audio generation on mobile devices utilizing Arm chips, allowing offline functionality with royalty-free content. The model can quickly generate audio from text prompts and will be showcased at Mobile World Congress 2023 in Barcelona.
Key points
- Stability AI partnered with Arm to optimize audio generation for mobile.
- Stable Audio Open generates audio from text descriptions.
- The model avoids IP risks by using only royalty-free audio for training.
- Performance improvements show generation speed increased by 30x.
- Currently not available for download but aims for future accessibility.
- Stability AI faced financial difficulties leading to management changes.
Why it matters
- Innovation in audio generation aligns with IT trends of offline AI capabilities.
- Represents a shift towards locally processed AI models due to privacy concerns.
- Reflects the growing popularity of generative AI across creative industries.
The collaboration between Stability AI and Arm marks a significant step in making powerful audio generation accessible on mobile devices, showcasing a trend towards efficient offline AI solutions amid past challenges facing the company.