so i searched google scholar for audio source separation with public code and found a less recent paper that’s already had its source uploaded i think at https://arxiv.org/abs/2305.07447 https://github.com/bytedance/uss
uss is short for “universal source separation”. the say it’s actually on python’s pip! “pip install uss” . this looks a little promising in that it’s an installable framework, wow. it seems a little nuts and bolts though, it looks like they might expect you to learn audio query encoding and such, which is doable i imagine.
