I’m using the version of spectral layers that’s include with Cubase 11 pro to separate vocals & music. The vocal stems still have some of the music in parts so it looks like the algorithm doesn’t work so well. I know just how complex this is but wondered if there are any parameters that I can tweak to minimise this ? The sensitivity option makes it worse the higher I set the value so is there anything else I can do ?
The results you get will always depend on the source material. It’s just the nature of the beast.
You could always try to clean the vocal layer a bit further with the Voice Denoiser (under the “Process” menu).
I don’t know if your version (“One”) has this function, but in “Pro” you can tweak the sensitivity when choosing “Unmix stems”. I often use the sensitivity maxed to “4” for the vocals stem.
Ok, that’s great Ill give it a try, thanks for the advice 
I dont understand what you mean by “4”. Do you mean 4 stems? Because the sensitivity only goes from -1 to 1 (there is no “4”)
That’s right, I meant ”1”, not ”4”.
There are sensitivity sliders in other dialogue windows that go to ”4” (e.g. in the ”Voice Denoiser” window, try that one!) but not in the ”Unmix Stems” , I mixed them up. Sorry.