Saw this today and thought it might be interesting to the community here:
https://github.com/deezer/spleeter
Some discussion here:
https://news.ycombinator.com/item?id=21431071
Apparently they are having a lot of success with these trained models to extract voice, piano, guitar, drums etc from mixed tracks. The results sound fantastic and seems to be a promising project.
https://github.com/deezer/spleeter
Some discussion here:
https://news.ycombinator.com/item?id=21431071
Apparently they are having a lot of success with these trained models to extract voice, piano, guitar, drums etc from mixed tracks. The results sound fantastic and seems to be a promising project.