It is VERY difficult, if not impossible, to remove the vocal from a recorded track, as all the audio has been compressed to one single audio track. There are some programs able to remove vocals, but the quality will almost never be satisfactory. Often you will get a result where some vocals have been removed but also som instruments!... So if you wish to remove the vocals from a professional track (from CD, LP, MP3 etc.) then I would suggest that you look for a accapella version of the particular track instead!
If you still wish to pursue your original i thought I have heard that Wavelab 4.0 has a function called "Voice Attenuator", dedicated to remove voice/vocals. It should be decent, but will probably not do a perfect job.
The hard way is to remove stereo or mono sound from a track and then subsequently remove frequencies which are not similar to the vocal.....
But it is difficult and time-consuming.
Best solution would be to find a accapella version.
Removing vocals (not going into copyright here!) can be done fairly easily if the original recording was made in stereo and the vocalist was "centered" and the rest of the group was "off center axis" at least somewhat. If you have a "mono" file, give up before you even attempt this.
First of all get Goldwave from http://www.goldwave.com
It is shareware but it will allow you to do what need with in the limitations of the shareware. (Buy it if you can afford it around $50. US and it's worth it!)
There is a "one button" remove vocal feature with various settings. It's works by performing a basic a phase cancel. Some settings work better than others, you can preview them all.
Be aware that, as some people stated here, you can rarely remove all of the vocal tracks but you can usually make it usable for live performances, etc. One of the big issues with completely removing thevocal is that the "reverb" applied, post production, to a track is usually "off axis". This means that it won't phase cancel.... so what you may actually be hearing (after the removal) is not the vocal track but the "left over" reverb...
One of the things that I do, that really works well for me, is to run both a "high pass filter" set at a frequency over 11,000 Hz and save whatever is left to a file. Then I run a "low pass" filter and save what is left to another file. Then I run the "vocal remover" and save that to a file. Finally, I merge them all back together and save them to a new file.
By doing this you are able to maintain (add back) some of the lows and the high frequencies that might have been removed from the vocal elimination. Again, you will have to adjust the high pass and low pass according the frequency range of the singer, so there is no simple solution.