Pretty much every software that extracts stems from music files is targeted to regular music, rock, pop, jazz, rap, etc. The instruments that the Skynet/Genisys algorithm looks for are a drum kit, guitars, bass, piano, etc.
Is it fair to assume that the Pro version of Spectralayers is the same? Has anyone tried it for music like film scores? Think Hans Zimmer, Michael Giacchino, etc. Of course I don’t expect any software in the next 5-10 years to isolate each instrument of an orchestra because that’s way too much, but can Spectralayers Pro at least isolate Woodwinds, Brass, Strings and Percussion?
1 Like
The problem is there are no specific models trained for some individual instruments, at least on high quality. There are number of high quality open source models, but most people seems to concentrate only karaoke…
I hope Steinberg trains their own models
1 Like
You hit the nail in the head. But I would say that problem is derivative of the first problem, which is, there’s just no interest. I spent hours searching online for software or services that provide stem extraction for orchestral music and found nothing.
It would be great if people who are not coders could have access to some website to upload audio files to train those models, that way we could upload examples of the most typical orchestral instruments, both individually and in groups, like strings, woodwinds, brass and percussion, so that eventually we can have the same for orchestral music.
2 Likes
You could try UVR’s discord channel. I think it’s called audio separation. There are some hitech gurus and coders, which are really nice and helpful people. They also train models and help others to do so too.
1 Like
Thanks, I did go there the other day and got in touch with someone who’s going to help with this. Hopefully some day those tools will work for orchestral music.