Ay! Aye!

[JGale]

Introducing Diff-A-Riff: a new AI model that create instrumental accompaniments for your music. It uses a pretrained Autoencoder (CAE) and a Latent Diffusion Model (LDM) to generate music that fits your chosen context. With Diff-A-Riff, you can provide a musical context or a text prompt to guide the model in producing a fitting instrumental track. Whether you're a musician looking for inspiration or just curious about AI in music, Diff-A-Riff opens up exciting new possibilities!

https://sonycslparis.github.io/diffariff-companion/

[Steve Haynie]

Maybe AI can start producing software to put the people who created this crap out of business.

[hamerhead]

Not so fast, Steve! I always come up with 'great' ideas for songs, but I'm a shit songwriter. If you could feed AI the parameters - the 'great' ideas - and have it spit out as a completed work, and then bring it to your band to learn, tweak and perform it, why not?

[velorush]

At first blush, I have a little bit different take on this. I have been in a musical vacuum for more than ten years. I've had no musicians local to me with whom to interact.  There are no gigs, no jam sessions, nothing.

I'd say I'm hopefully interested.

[Nan0brain]

Interesting.

Reading the introductory disclaimer in the video, maybe it makes more sense as to why they recently chose to buy the Queen-flavored dataset.