- Stability AI released Stable Audio Open, an AI model for generating sounds and songs trained on royalty-free recordings.
- The model can create drum beats, instrument riffs, and ambient noises, but lacks capabilities for full songs, melodies, or vocals.
- Stable Audio Open is not for commercial use and has biases in its training data, which lacks cultural diversity.
OUR TAKE
While Stable Audio Open’s launch can cement its parent company’s status in its sector and provide relevant services for its users, the company’s use of training data may infringe on its owners’ copyright and result in a series of legal issues.
–Audrey Huang, BTW reporter
Stability AI has introduced Stable Audio Open, an AI-powered sound generator that uses text prompts to create audio clips as long as 47 seconds. Trained exclusively on royalty-free data, it can produce various audio elements but is limited in creating full songs or vocals. The model’s use is limited to non-commercial purposes and shows biases due to its training data. This release aims to address controversies around AI and copyright while promoting Stability AI’s premium services.
Stability AI’s sound generator
Stability AI, known for its art generator Stable Diffusion, has launched Stable Audio Open. This AI model generates sounds and songs based on text descriptions, using over 486,000 samples from free music libraries. It can create various audio elements useful for video and film production but is limited in producing full songs or vocals.
Also read: The Next Music Revolution: AI Sound Effect Generator
Also read: Powerful rise of AI in film celebrated at LA festival
Stable Audio Open’s limitations and biases
Stable Audio Open cannot create full songs, melodies, or vocals. What is more, the model has biases from its training data, showing its lack of representation of diverse cultures. It is also restricted for non-commercial use, highlighting the challenges in balancing AI capabilities with ethical and legal considerations.
Dealing with copyright concerns
The release of Stable Audio Open comes amid growing concerns about AI and copyright. Stability AI’s move to use only royalty-free data for training aims to address these issues, setting it apart from practices that some argue infringe on copyright. This step also promotes the company’s premium services, enhancing Stability AI’s position in music.
