I've been spinning since 1990 and have been through all sorts of genres. However for majority of the time I have been heavily into Drum n Bass and Chicago House and those are my main stay of music. However I like all music which is outside mainstream and pop. I've also produced for many years on and off since 1996. Recently since late 2008 I have created a record label called Mixtura Records with my partner in crime Smokingroove. www.mixturarecords.com www.pazngroove.com However my career is not producing or DJing. I have no patience for the professional DJ world as I make much more money as a Technical Director for Flip Media which is the largest online interactive agency in the Middle East. I currently reside in Dubai with my wife Natasha, although we hail from Toronto, Canada and have also lived in Amsterdam, NL for 8 years before progressing to the hot arid concrete jungle of Dubai.

Mon 04 Feb 08 @ 12:15 pm

A dude approached me by PM to chat abit about removing vocals. It's not an easy task, so I thought I'd start this thread to show what I know and to start an exchange. Some precursor knowledge, I have DJ'd about 17 years and produced for 10 on and off. Not trained in it, but have access to alot of studios and know alot of people who do it professionally. The following is some data I've picked up on the subject over the years:

Well vocal separation is no easy task. Quite often the remixes you hear (if its a good one) the vocal track was supplied by the artist.

If you need to remove the vocal, this is a whole different story and many producers have their own techniques.

Just to say immediately, you cannot remove vocals on the fly, it's an arduous process you need to do before hand.

The easiest and quickest way is to use filters and paragraphic eqs. Vocals generally tend to be in the range of 1k-5khz in frequency, so you can use two filters to completely drop the lows and the highs out.

The you use a paragraphic eq to 'notch' out certain frequencies of instruments that exist in the same range of the vocal.

There is one other method which I dont use much, but sometimes can be effective, which is to do the reverse of above. You use only a paragraphic eq to remove the vocal and not the instruments, since you getting rid of less portions of the track (only one instrument; the voice). After this, you take the result and invert the wave. Then you mix down the inverted wave over the source. If this is done properly, you will in theory cancel out the instruments and leave a cleaner result of the vocal which you can clean up with some more filters and paragraphic eqs.

Long story short, its a trial and error process that you need to mess about with for a long time. The result is also relative to the source material, as in the quality of result depends on the instruments and composition of the source. Saxaphones tend to compete in the vocal range, and other instruments may. It all depends on the track you're trying to remove.


Sorry, this isnt really a straight answer, because their isnt one. These kind of techniques require alot of post-production and require some knowledge of sound. But you can learn yourself if you spend the time to do it. I suggest having the correct tools as well. Doing it in vdj wont cut it. You need a proper editting tool like Steinberg WaveLab or Sony SoundForge (used to be by Sonic Foundry). With that, you need something like the Waves Gold Bundle VST/DX plugins which come with a quality paragraphic eq.

Essentially a paragraphic EQ allows you to manipulate multiple bands and specify which frequency to modify and how sharp, like taking a big or small slice of a pie. Only with this tool can you zoom in on the correct range. AFter that you can use the onboard spectral analyzer to visually see (before and after the eq) what frequencies the vocals exist in so your work is a bit less trial and error....