Stability AI, a London-based startup known for its open-source image-generating AI model Stable Diffusion, recently unveiled Stable Audio, a tool capable of creating "high-quality" 44.1 kHz music for commercial use using latent diffusion techniques. This development follows the earlier release of Dance Diffusion, Stability AI's first venture into generative audio. Harmonai, an organization funded by Stability, initially created Dance Diffusion, but it was not widely adopted due to the lack of a user-friendly interface. Stable Audio, with its 1.2-billion-parameter model, offers more control over synthesized audio content, allowing users to specify duration and style through text prompts.

While Stable Audio shows promise, some concerns surround its use and potential copyright implications. Stability offers subscription tiers for users, with limits on the number and length of tracks generated. However, questions remain about how artists whose work contributed to training the model will be compensated, as the terms of the deal between Stability AI and its music library partner, AudioSparx, are not fully disclosed.

‍