TLDR: Stability AI has unveiled Stable Audio 2.5, an advanced audio generation model specifically designed for enterprise use. This new AI tool aims to empower brands to create unique and high-quality ‘sonic identities’ for marketing and various channels, addressing the current low adoption of custom audio in branding despite its significant impact on brand memorability and engagement.
London, UK – September 10, 2025 – Stability AI, the company behind the popular Stable Diffusion image generator, has announced the launch of Stable Audio 2.5, a new audio generation model tailored for enterprise-grade sound production. This release, unveiled on Wednesday, September 10th, marks a significant step towards enabling businesses to develop distinct ‘sonic identities’ across their various platforms and campaigns.
The model is designed to tackle a notable gap in corporate marketing strategies. Citing Ipsos research, Stability AI highlighted that while custom audio can make brands up to eight times more memorable, only a mere 6% of creative campaigns currently incorporate a sound identity. Further internal research from Stability AI indicates that a substantial 86% of brand engagement is influenced by audio, yet few brands leverage custom audio effectively.
Stable Audio 2.5 boasts several key enhancements aimed at professional creative teams. It offers ‘faster generation, smarter composition, and enhanced workflows.’ The model is capable of generating three-minute compositions in under two seconds when utilizing H100 GPUs. It is optimized for music, producing dynamic and ‘richer multi-part compositions’ that can include intros, development sections, and outros. The system is also highly responsive to nuanced prompts, allowing users to specify emotional tones like ‘uplifting’ or technical elements such as ‘lush synthesizers.’ A new ‘audio inpainting’ feature further enhances creative control, enabling users to upload existing audio files and specify areas for the AI to generate additional content, seamlessly extending tracks based on context.
In a statement addressing industry concerns, Stability AI emphasized that Stable Audio 2.5 is ‘commercially safe and trained on a fully licensed dataset.’ This assurance comes amidst ongoing discussions and lawsuits regarding the use of copyrighted material in AI model training, with Stability AI itself facing legal challenges from entities like Getty Images and a class-action lawsuit from illustrators. The company’s terms of service explicitly prohibit copyrighted material uploads, and content recognition systems are in place to monitor compliance.
To facilitate widespread enterprise adoption, Stability AI has forged a partnership with Amp, a WPP-owned sound branding agency. This collaboration will see Amp and Stability AI co-develop enterprise solutions, making Stable Audio 2.5 accessible to WPP’s extensive global client portfolio through the advertising giant’s technology platform, WPP Open.
Also Read:
- Deepdub Unveils Groundbreaking Real-Time AI Voice Model for Live Global Broadcasts
- PicLumen AI Introduces Primo and Picflow: Advanced Models Revolutionizing Creative Content Generation
Stable Audio 2.5 is available for immediate use via StableAudio.com, the Stability AI API, and through partner platforms such as fal, Replicate, and ComfyUI. Enterprises also have the option for on-premises deployment with an enterprise license, offering flexible integration to meet diverse operational needs.
