22 - Equalization





We have already introduced the fundamental concept of frequencies. A very useful fact is that you can break any sound into different sine waves at different frequencies, which in turn makes it possible to sculpt these frequencies and change the way we hear a sound. In real life different frequencies are boosted and others are lowered in various contexts: if I speak directly into your ear the frequency content is not the same as if I speak 2 meters away from you or if I hold a pillow in front of my mouth, even if the sound that comes out of my mouth is exactly the same. In between the sound producer (my mouth) and the sound receiver (your ear) the frequency content changes. This is one of the main ways we can figure out what kind of space we are in, how far away a sounds is, if it is in front of us or behind us and many, many other things.

With signal processing we can do the same thing with filters. Instead of holding a pillow in front of my mouth I can use a lowpass-filter to remove most of the higher frequencies. I can fake the sound of blowing in a bottle by running white noise through a band-pass filter, so that only a few frequencies come through. I can make a recording sound like it’s coming from a tiny speaker by removing the lower frequencies. And so on. Let’s take a look at the frequency spectrum of a couple of different filters. First of all pure white noise with no filters applied:

Next, the same sound with a lowpass-filter applied at 1200 hz:

Now, with a lowpass-filter at 200hz:

The highpass-filter works in the opposite direction. Here at 200 hz:

At 1200 hz:

Let’s do this in Audacity, so you can hear the difference. Open a new project and call it Filters.aup. Generate 10 seconds of white noise. Listen carefully to the sound. Select the resulting clip and go to Effect -> Low-pass Filter. This might be in a slightly different location depending on your installation. Set Frequency to 1200, keep Roll-off at 6.dB and click OK. Listen to the difference. Type Ctrl-Z to get back to the unprocessed version. Undo the undo by typing Ctrl-Shift-Z. Go back and forth between the filtered and the unfiltered version to understand how the sound is affected.

Mute the track, and create a new track with another 10 seconds of white noise. Run the low-pass filter on this as well, but this time set Frequency to 200 hz. Listen to the result. Swap back and forth between the 1200 hz and the 200 hz versions.

Mute the tracks, and repeat the same exercise with high-pass filter this time.

Combining several filters into one interface is a very common form of processing, and is called equalization. For a more intuitive filtering experience you can use the Filtercurve effect, which allows you to shape the frequency content of a sound with an arbitrary number of filters. Just click on the curve where you want a new point and drag it around. If you want to get rid of a point drag it out of the window. Click the Preview button to hear the result.