The launch comes as Stability AI faces pressure from investors to generate revenue from its $100 million capital. The company has partnered with commercial music library AudioSparx to train Stable Audio using around 800,000 songs from independent artists. Users can monetize their works but not necessarily copyright them. The company has faced criticism for training models on artists' work without compensation or consent, but it has an opt-out mechanism for artists who do not wish their work to be used in training data.
Key takeaways:
- Stability AI, a London-based startup, has released Stable Audio, a tool capable of creating high-quality music for commercial use using a technique called latent diffusion.
- The tool was trained on a collection of around 800,000 songs from the commercial music library AudioSparx, and can generate songs and sound effects based on text descriptions.
- Stability AI has faced criticism for not compensating or informing artists whose work was used to train its models, but has implemented an opt-out mechanism for artists who do not wish their work to be used.
- The company is under pressure to generate revenue from its $100 million in capital, amid reports of financial mismanagement and a high burn rate.