OpenAI said to be working on a music generation tool

The Information (paywalled) is reporting that OpenAI is well underway to create a music generator.

It could be used to generate music from a text or audio prompt — or make a guitar background to a vocal track, or add music to videos.

Recruiting at The Juilliard School
Apparently, the AI lab has partnered with students from The Juilliard School of music to annotate scores that could be used to train the generator.

The work doesn’t seem particularly advanced beyond this point, TechCrunch and Engadget reports.

It is also unknown whether it will be a a standalone app or integrated into other offerings, like the video app Sora 2 or more broadly into ChatGPT.

Booming AI music market
The AI music business is on a tear lately, with Suno reportedly passing the Turing test, meaning people can’t distinguish between generated music and that produced by humans.

Also, the AI Avatar/artist Xania Monet, which is the brainchild of Talisha Jones, signed a $3 million record deal after reaching the top 10 of Billboard’s R&B Digital Song list.

The OpenAI music generator, though, seems in early development and I wouldn’t expect any product any time soon — and by then the market may be booming even more.

Read more: Original reporting by The Information, writeups on TechCrunch and Engadget. Suno discussion on r/Singularity.