It's very hit and miss but, not sure what models you use, but this is how I do it.
Run my file through BS Roformer and get an instrumental and vocal separation.
Then run the vocal separation through UVR BVE Back Vocals model, this tends to work most of the time.
If it doesn't there is now a MVSep Male/Female model that I try.
If at this point it doesn't work or is not good enough for what I want I give up.
Hopefully this answers your question, if you have anymore post them and I will get to them eventually