Sound suppression technique

How to remove a mixed sound among others?

Sound suppression technique
For example, you want to remove a singer's voice from a piece of music or just keep that voice; or delete the kick that you should not have saved during your last composition.
I have a very simple technique, which takes little time to set up and which is much better in terms of quality than using filters.
This is called selective anti-phase. We are going to put a wave in anti-phase with respect to another to obtain a remainder.




1. Before starting ...

But you must respect certain conditions:

  • 1. It is necessary to have the sound to be subtracted (the one which contains the sound among the others).
  • 2. You need of course the subtraction sound, the subtracter (the one you want to remove from the other sound). It must be exactly the same as the one you hear (same amplitude, frequency, time, etc ...) otherwise it will only add sound.
  • 3. A software like [Reaper], Audacity or Sound Forge, but having the capacity to invert the amplitude and to mix at least 2 tracks.

2. The theory

The principle of this subtraction is simple: we will reverse the amplitude of the subtracting sound to make it subtract from the other sound.
Example:
I created an FM synthesis made up of 2 waveforms: Saw (Sawtooth) and Square (Square) of the same amplitude and with the frequency of 250 Hz (Hertz) and 500 Hz respectively, all for 2 seconds. Here you only see a fraction of a second:
Sound suppression technique




Here I want to remove the square sound of 500 Hz for example.
Here is my 500 Hz square sound:
Sound suppression technique

To subtract it, you have to invert its amplitude:
Sound suppression technique

And mix it with the other sound, by putting it exactly at the same time of the sound to be subtracted inside the sound having the 2 waveforms.
So the mixer will add the amplitudes and as (-9,4) + 9,4 = 0, the square sound will disappear to make way for the sawtooth sound:
Sound suppression technique

It's complicated, I admit, but practice, as you will see, is easier.

3. The practice

From there, we can start.

We are going to get the vocal of a song.
For that we will have to get the instrumental and the vocal mix of the title.
(Note that you could have stupidly asked the author for the vocal or else bought it, it's safer, legal and faster).
The structure of the two versions must be IDENTICAL as well as the digital quality.
Then pray that the mastering didn't make too much of a difference.

We're going to use Cockos' Reaper to do the thing.
Go look for it here at the speed of a race! -> https://www.reaper.fm/download.php
This software is very efficient, cheap, free 30 days with no function removed. Pass those 30 days and you will just be allowed to be blocked for 5 seconds.


Okay, you're smart enough to set it up on your own. Open a new project with.
Then go to the menu Insert -> Media File.
Go find your two versions.



There you have it with two great tracks and a sound that saturates, but it's not too bad.
Put a track in anti-phase.
Like this: Sound suppression technique



And with any luck you should be done with the vocal (and some static).
If that doesn't work, check to make sure the waveforms are seated properly.

Wedge them as close as possible to the connection. (SHIFT + Left click to move freely).
You will never be able to position them correctly like that with the mouse.
This is why you are going to add a JS effect which will help us.
So click on the button fx of the track which is clearly late.
Click on Add -> JS -> Delay / time_adjustment and do Ok.
Your effect opens.
Check that delay amount (ms) as well as wet mix (dB) is at 0, then check that dry mix (dB) is at -120.
You just have to modify the last parameter (additional delay amount) using the keyboard.
This will create a very precise delay or advance on the sound of the track and so you can stall the two monsters quietly.
Little tip: The more the bass disappear and the treble remains, the closer you get to the point.
Go forward 10 by 10 samples, until you have a fairly attenuated sound.
Then, refine by making 1 by 1 sample. And so on ... until you find the right value (Most importantly, stay calm. It could take 2 seconds or it could take an hour)


If you are still not satisfied with the result, then go to the filters, otherwise ask the music author.

Good luck!

add a comment of Sound suppression technique
Comment sent successfully! We will review it in the next few hours.