Our single most common request at Blaze Audio is for a tool to remove vocals from pre-recorded music. Removing vocals or any single instrument from non-MIDI music is much more difficult than it might seem. Each voice or instrument becomes inextricably mixed into the audio data, much like flour after it is cooked into a cake.

The difficulty of the task didn't daunt our tireless programmers. They have been researching the best techniques available for reducing or removing the vocals (or any other instrument) from a song and have put together a package of tools that can be highly effective. This page will describe how the package works and give you some hints on how to use it.

If you add an inverted copy of the right channel to the left channel you eliminate everything in the center channel. This approach is used to remove the vocals from recordings that have the vocals in the center channel. It can also remove drums or anything else in the center channel, and it leaves you with a mono result, but it works quite well on some recordings.

There is a technique for minimizing the loss of non-vocal sounds in the center channel. Ethan Winer describes it quite well in the following article:

http://www.ethanwiner.com/novocals.html

Please note that his technique (described at the end of the article) for producing a file that still has the original stereo imaging is flawed. He noticed that it doesn't sound good; if you think about it there is a good reason it doesn't sound good.

There is a way to take all or most of the vocals out of any piece of music whether they are in the center channel or not. You can simply use an equalizer to reduce the amount of energy in the band occupied by the voice. It takes a fair amount of work, and it will also remove other sounds in that band, but it can be quite effective.

This technique works best with a very narrow filter set to the correct frequency for each note of the vocal track. That's time consuming, but worthwhile for some users.

It's worth noting that the Blaze Audio Vocal Reducer can be used to reduce or eliminate any instrument or voice from a recording. It will be easier to get results quickly if the sound you want to remove is in the center channel, and the results will be better if the instrument's sound is limited to a fairly narrow range, but these tools will work for any range of audio.

To summarize, there are two basic techniques that can be used to remove vocals- center channel suppression and selective, narrow band equalization. In addition, the fact that the center channel suppression technique renders the file to mono makes it important that we provide some way to convert the mono file to a reasonable simulation of stereo.

The first thing to do is to determine whether the vocals are in the center channel. To do that, load the file to which you want to apply Vocal Reduction, click on Vocal Reducer in the Tools menu, and then click on the button labeled "See if Center Channel Removal will help this recording". It will take a few moments to process the file, and then it will start to play.

Note that the trial version will only work without limitation for 7 sessions of 30 minutes each. During the 7 trial runs it is fully functional.

If you want to experiment with reducing vocals later in the track you will have to use the editing tools to cut out the portion before the spot you want to work on or purchase the full version. Please check our Editing Tutorial for information on editing audio files.

Are the vocals gone or significantly reduced? If so, you're in luck, and you can do this the easy way. If there isn't much change in the vocals skip ahead to the section on the Note Scrubber. If you do hear a useful reduction in the part you are trying to eliminate, click on "Yes" or go directly to the Center Channel Removal effect.

Usually the center channel has more than just the voice- often it has the drums, bass, and even other instruments. So you don't want to remove any more than you have to. Use the Top and Bottom sliders to set the top and bottom of the range to be eliminated. The presets will help you get close. We use musical notation indicating a range from C in the bottom octave (C1) to C in the top octave (C8).

You will probably need to do some experimenting to find the best range to eliminate. Pick one of the defaults or set the sliders to your best guess, then click on "Remove Center Channel".

Listen closely to the results- is all of the vocal removed, but very little of the other material? If not, decide whether you need to increase the range or decrease it, and whether the top, the bottom, or both need to be changed. Keep trying until you like the result, then click on "Save the Result".

Once you have removed the right part of the center channel, you need to convert the result to stereo in the next step.

In mono to stereo conversion we add a very tiny offset to one of the channels, and then apply equalization differently to each channel. It's so simple and effective it seems like magic!

Most of the time you will get a good result with the default settings. If you want to, though, you can experiment with every part of the process- you can change the amount of delay and change the amount and kind of equalization applied to each channel. You can even use our other effects to do more processing later. For example, you can add just a little echo to each channel, but make the echo different for each one.

Depending on how much work you put into it, the Note Scrubber will allow you to remove almost any pitch from a recording. The only exceptions will be drum beats that usually cover so many pitches that removing them will remove all other sound at that spot. We use a similar filter in our Notch Filter for getting rid of 50 or 60 cycle hum. In fact, if you have a specific pitch you really need to eliminate, the Notch Filter removes everything at a frequency, whereas the Note Scrubber allows you to adjust the amount you remove.

The easiest way to use the Note Scrubber is to apply it to the range the voice or instrument is in over the entire song. If the voice is on one side or the other it may be best to Scrub just that channel. You can set the level of Scrubbing with the Scrub Depth slider.

However, it is very likely that there are other sounds at the same pitch. Once you find the right range for the voice, you can select the areas with singing, and apply Note Scrubber just to those areas. That way sounds in the vocal range that happen when the voice is silent are not affected.

If you are really determined to get the maximum voice reduction, you can go through the song one note at a time, changing the pitch to be removed each time so that you only get rid of the one note. It's a lot of work, but it is very effective for that song you really want to get right!