Stable Audio

I tried it out and liked it a lot. That the produced audio is at 44100 kHz creates a lot of possibilities. I’m not a music expert, so take this with a grain of salt, but here are some of the uses I see for stableaudio:

- Generating samples for sampling VST instruments.

- Generating melodies.

- Generating particular drumkits, which can then be re-timed and quantized in a DAW.

Good prompts will require knowledge of music genres and musical theory, and then a music producer would probably want to do all sort of things to the audio to get a proper song. For example, extract musical notes, fine-tune the tempo, etc. All said, I’m not sure this would be a huge time-saver for any serious musician, but it’s definitely going to be a power tool.

Thanks for sharing!

Stable Audio is a generative AI tool for music & sound effect creation from Stability AI.

It lets you enter a text prompt and a duration, and generates 44.1 kHz stereo output. It uses a latent diffusion for audio model, a similar technique to what’s used for image generation in Stable Diffusion.

We’d love for you to give it a go! Here are some prompts we like:

A solo bass guitar stem, funk, groove, 112 BPM lo-fi hip hop beat, chillhop drum solo

We’ve found it’s good at EDM, ambient music, and combining different ideas, and not so good at jazz, classical, and other more melodic genres. We’ve prompted the model a lot, but we’re a small team - if you try it out and find prompts that work particularly well, we’d love to hear them.

I also really like http://www.word.band. Has an option to share and see other AI music on it as well.