or another way to do it? Thanks.

Sonar as a DAW doesn't lend itself to presets. None do.  X2 came with R-Mix which can do a lot to isolate a voice from music. But it will not be easy.  EQ will get rid of frequencies  above and below the frequency of the voice but again those frequencies at the same voice frequencies will be hard to remove.

There is no preset for that function, and whether you can do it at all with anything other than a heavy-duty program designed for forensic analysis depends on how much isolation you require, whether the voice is mono or stereo, whether the background music is stereo or mono, the levels of the two, the frequency ranges of the music and voice, etc.
 
The closest you can come with something "normal" would be something like Sound Forge's noise reduction tool. Select spaces where there are no vocals and make that your "noiseprint." SF will subtract sounds with that characteristic from the file. You'll have to do this lots of times and it will likely affect the voice to some degree. Then you can filter out everything except the vocal range. It won't be perfect, though.
 
Removing vocals is much easier.
 
 

Sony SpectraLayers is a high end solution. I don't see how anything else could do any better.
http://www.sonycreativesoftware.com/spectralayerspro?keycode=64588
 

It's a tough one, for sure, and I have not run across anything that can do it.
 
Perhaps the one from Sony mentioned above can get you close.
 
Bob Bone
 

dougalex
Sony SpectraLayers is a high end solution. I don't see how anything else could any better.
http://www.sonycreativesoftware.com/spectralayerspro?keycode=64588

You can do it in SpectraLayers but plan on spending some time getting the result you want. There is still no way to unmix something without artifacts. SpectaLayers is a great tool for removing things like cell phone rings, pops and hisses, etc but It would take you days to weeks to unmix even the simplest song and even then it's not going to sound right.
 

good to know. Thanks for all the input!

IMHO RX does a better job at isolating sounds than spectral layers. Maybe I did not dive in deeply enough but the spectral layers demo did not do much more for me than the specttal tools in Audition.

Possibly stating the obvious here but you should be able to filter out a lot of the stereo content (e.g. a lot of the music) using the channel tools and phase cancellation. Basically you would first get a version where you phase cancel one side which should make the vocal almost disappear. Then you put that version against the original and invert the phase of one of the two, leaving you with only the centered content (all of the vocal and hopefully a LOT less of the music). Probably most of the bass will be left but a really steep HPF should take care of a lot of that and after that you could still use noise reduction software.

If you have the backing track without vocals you can phase subtract.