Just wanted to mention another (free!) tool that I've been trying out for the last few days. About a year ago someone mentioned MVSEP, but with no explanation and I had forgotten about it.
https://mvsep.comThis one isn't software to download and run on your own computer, but a website where you upload your file and then it gives you split files which you can download. It seems like it would be slow, and with a free account you do have to wait in line, but it's been working for me in less than a minute. This would particularly be beneficial for someone who doesn't like to have many programs on their computer or has a slow processor.
I haven't yet tried the models to split up instruments, but I have used the BS Roformer model to split vocals (and crowdnoise) from music , and the MVSep Crowd removal to remove crowd noise which is at the same time as singing. Both worked very well, but not perfectly in some instances. Even if you're happy with what you currently use, this would be another one to try when you're not getting the results you want from UVR5, SpectraLayers 11, etc. Each different processor seems to work better than others in some instances but worse in some of them.
One advantage over UVR5 is that MVSEP can handle sample rates other than 44.1, although you have to be a paid member to use that feature.