Lossless stems

saradis

Yes, you got my point, full ALAC support would be great, but depending on the source material quality and genre it's not worth the space (for those on macbooks at least) sometimes because the separation alone makes it far worse compared to the source than the AAC compression. Some overproduced pop songs sound terrible no matter the stem quality, it's like they get disjointed 😛 I'll be sure to try audioshake myself

Demus

The stems separation algorithm of all the different software is almost the same depending on the file type quality and the genre of the song. Serato, Virtual DJ, Traktor DJ Pro 4 and Djay Pro 5 which uses audioshake the Difference is real-time verses pre-analyzed. as @saradis said "ALAC support would be great, but depending on the source material quality and genre it's not worth the space (for those on MacBook's at least) sometimes because the separation alone makes it far worse compared to the source than the AAC compression. Some overproduced pop songs sound terrible no matter the stem quality, it's like they get disjointed" Let see how much better the algorithms will get as time goes by.

Nico_Tf

By the way, I’ve looked around, and here’s what seems to be the current state of the art. They’re cheating a bit, as they process songs by hand, yet they developed their own models. Really quite nice, give it a listen.

https://ai-lyd.no/en/demo/toto-africa-ai-demo-33/

But Toto should be fairly easy, here’s Britney:

https://ai-lyd.no/en/demo/classic-track-demo-16-baby-one-more-time/

Simon A. Billington

https://community.native-instruments.com/discussion/comment/173905#Comment_173905

Yeah, sorry. Or I misread what you had written. I'm little dyslexic so the latter is quite possible.