I've been using iZotope Rx to cut out the frequencies that i don't need so i get the vocals in a pretty usable form.

I'm just wondering if there's another way or plugin to do this?
How you guys usually do it?

Or get a friend to record them singing it.

Side note: I haven't been on music forum yet that wouldn't have to deal with this particular question every other week or so.

- So technically best method is the phase inversion, but you need the exact acapella exported from the same project. Sometimes it is available, but a lot of times it isn't.
- Few companies including iZotope and Adobe attempted to automate FFT based extraction, but from what I've heard it works on some kinds of songs, doesn't work on others. Also if the track has a lot of high frequency information, brillance of the vocal will be probably quite s***ty.
- You can try to play with multiband gates or pattern based noise removals, but it probably won't help. You'd have to be really lucky.

So basically it's damn hard task to "unmix" audio. In a lot of cases totally physically impossible.

A trick that sometimes works if you can get hold of an instrumental of the track, is to phase invert the instrumental and then line it up with the original in your DAW. But it doesn't always work very well, on top of which some producers have got wise to it and deliberately phase-invert some part of the instrumental mix, so that it won't cancel out.

...or when the instrumental got mixed or mastered differently.