Hello again Aubry,

another way that might work is to use Ultimate Vocal Remover. the tool is free and offers a variety of AI models (that is the training information to teach the AI engine). And that it's doesn't mean its cheap. It is regarded as par with the likes of iZotope. The multiple offers available as an app or online do not differ much with regards to the engine, but with the training models used. UVR offers a whole lot of such models and you can even combine some. Picking the right one is trial and error for each new song. But you may succeed. I found it useful, but somehow there is a sync problem when loading the stems into my Daw (maybe related to different sample rates and bit sizes the respective Software uses internaly). So I prefer to stay with my "One Tool Only" approach and do everything within my DAW. Stem separation is a native part and Melodyne thightly integrated.

“Musicians”: UltraPak 2025
DAW: Studio One Pro +
"Instruments": VSTs and REs
Notation: Notion 6
Interface: NI Komplete Audio 6
Mic: RØDE NT1-A
Controller: Panorama P6/KAWAI VPC1/Atom
OS: Win 10 64 Pro