It’s great to be back launching a product on Product Hunt! I’m VP of Audio at Stability AI, and I’m really excited to be shipping Stable Audio today. 8 years ago, I launched the first AI music product on Product Hunt, Jukedeck. Jukedeck was a tool for video creators to make original music using AI, and it ended up being used to create 1.5 million tracks. At the time, it felt futuristic. But generative AI has rapidly accelerated since then, and today’s tech is leagues ahead of where we were back then. Stability AI is best-known for Stable Diffusion, the image generation model based on latent diffusion. Stable Audio uses that same approach - latent diffusion - but applied to audio generation. This means you can enter a text prompt and a duration, and Stable Audio will generate an original track, in 44.1 kHz stereo. Because it’s a text-to-audio system that generates raw audio, you get a high level of control and a lot of variation in output, rather than every track sounding the same. One use-case we’re really excited about is musicians using Stable Audio to generate samples to use in their own music production. We’ve had a number of testers using it this way, and it’s an area where generative AI really comes into its own - as inspiration and source material for larger works of human creation. This use-case is helped by the fact that you can specify instruments, beats per minute, and other variables just by entering them in the text prompt. But because of the range of output you can generate - even extending into sound effects - we hope people experiment with different use-cases, and we’re exciting to see what people come up with. The model was trained on a large catalogue from our partner AudioSparx, a leading stock music library, in a partnership that includes a revenue share between the two companies. AudioSparx has provided an opportunity for their musicians to share in the revenue generated by the model if they choose to participate. It’s free to get started, so we’d love to hear how you get on with it - we’re particularly interested to hear any prompting tips you come up with. We’ve included a user guide covering the tips and tricks we’ve been able to work out, but we’re a small team and we’re sure that all of you will find lots more hidden corners of the model and smart ways to prompt it. We hope you enjoy it!

Stable Audio

Create music & sound effects with AI

Create music & sound effects with AI