If you have access to the elements of the production it is very easy to simply mute the vocal layers in the mix, but this post is written for situations in which you have one stereo file of the song and nothing else. There are a couple of ways to deal with this but at best these techniques will only be able to muffle the vocals and cannot remove them completely. It’s almost impossible to completely remove all the vocals from the song without compromising on the quality of the rest of the music. This is because vocal removal techniques are usually based on the way the modern music is mixed. 

Stereo tracks have two separate channels, with the instruments and vocals spread across them. Most of the instruments will typically be pushed to one side or the other, while vocals are normally placed in the "centre channel". To isolate them, you'll split this center channel and phase invert one of them. Songs with a lot of effects on the vocals may be quite difficult to separate and invert. There may be a little echo of the vocals that will not be removed completely by this method. Background vocals may also not be removed completely.

First, import the song into an audio editing software like Audacity. You can also use high end software like Pro Tools, Ableton or Cubase, but if you aren’t familiar with these or don’t have them, its easiest to go ahead and download Audacity since it’s free. Split the stereo track into its own Left and Right channels. Select one of the channels and ‘invert’. Since they both have vocals embedded in the tracks, select either one. Double-click to select the entire track if you want to remove the vocals on the entire song.

Invert the channel: Once you've selected the track, invert it using the "Effect" function and selecting "Invert." The song might sound a little strange after you play it. After inverting, the track should sound like it’s coming from the sides, instead of the middle.

You should still be able to hear some of the vocals at a lower volume, but the process is not complete yet. You'll complete the effect when you bounce it back to mono. Combine the two stereo channels back into one channel. You should now have one combined track that should have less amplitude. That means the vocals will be out and the instrumentation will be usable. You may still faintly hear the voice of the original singer in the background in some cases but that’s just something unavoidable. This entire process can also be done in other software packages, though the steps might be different. 

There are also several websites which can do this online, such as www.vocalremoverpro.com, www.phonicmind.com, etc. The results may sometimes be better than doing it manually by the above mentioned technique. A few professional audio suites offer even more customisation options which can lead to better results.

Hope you found the post useful. Check out mysticalankar.com for more info and music production resources like sample packs, sound banks and templates to take your productions to the next level!