Learning the Basics of Compression and Understanding Over Compression
Written by Jeremy Quintin for The Omega Studios’ School
If you search for the phrase “how to compress a kick” or “when to use a compressor” on Google, you’ll find out real quick that everyone and their grandmother has an opinion on the matter. “Cut out those transients,” says one source, “no wait, keep them in!” says another. “Maybe try some parallel compression” says somebody on the DJ Mixmag forums. Who is right? Who is wrong?
Well, after a few months of reading the many sources and trying out the techniques on your own, you’ll find that the grandmothers of the world seem to be somewhat right and somewhat wrong. The issue is that the right way to use compression is entirely dependent on the incredibly specific scenario of dynamic balance which you find yourself up against. The thousands of online explanations for compression out there cannot account for your specific auditory needs, nor can they read your mind. In other words, an article on compressing a kick is not going to explain how much to compress your kick, where and when to compress your kick, or even what purpose that compression is going to serve. Worse yet, it might assume you’re compressing your kick within a mix of several other sound types that you never use, or for a specific genre that you never compose in.
But you didn’t come here to listen to me wax poetic on the limitations and obscurities of language and mindreading. You came here because the title of this post is “Attack of the Compressors!” You read that and figured I might shed some light on the nuances of compression. Meanwhile, all the audio engineers are groaning at me. Oh come on, enjoy a pun once in a while. Be thankful it wasn’t “Take it to the Limit-er!”
If you’re a sound designer, audio engineer, musician, etc., the only way to know what kind of compression process your sound needs is to understand how the very process of compression works.
So what is compression? Compression is quickly timed adjustments to the amplitude of a sound, based on its input level, to keep its overall dynamics consistent. That’s it, that’s all there is to it. And yes, in theory, you can create the same effect by moving the fader on your track down and up really fast with the audio signal.
So why compress when you can move the fader? Well I guarantee you can’t move the fader with the speed or precision that a compressor can. When I say “quickly timed adjustments,” I’m talking about milliseconds of response time. A sound that goes into a compressor is monitored and so quickly adjusted that a compressor can respond to volume changes before you even realize they’re changing.
There are two sides of audio to compressors and four big functions that instruct the compressor on how to actually compress (word of warning: these functions may vary by compressor, and many compressors include additional features. But these make a good starting point). The first side is the input, which is what your signal sounds like before compression. The second side is the output, which is the sound afterwards. Hey look at that, it’s the same as every effect ever!
Once the signal is input into the compressor, the compressor needs some instructions on how to treat the signal, or else it won’t do anything. The first instruction to be aware of is the threshold. The threshold, marked in decibels, is the point above which the compressor introduces level adjustment. Below the threshold level, the compressor doesn’t actually do anything but pass the signal through unchanged. Let’s say for example you set a threshold at -10 db. Now, every time the input signal gets up to the -10 db, the compressor becomes active. Every time the signal drops back below -10 db, the compressor stops compressing.
However, without instructions on how much to compress, the compressor still won’t do anything, even with the signal past the threshold. In order to actually make compression occur on the signal past the threshold, you have to adjust the ratio. In this case, the ratio is between how many decibels go in and how many decibels come out. Simply put how much the output is allowed to increase based on how loud the input is. If you have a ratio of 5:1, then for every five decibels of increase over threshold that go in, one decibel comes out. Note that five decibels is not a value to be met before sound goes out, but a guide by which we can predict the level that comes out. In other words, if four decibels go into a 5:1 ratio, 80% or four fifths of a decibel comes out.
Where does it come out? Why, to the output connection of course. At this point we haven’t learned every function of a compressor, but it’s enough to finally put our compression scenario to the test! First, let’s make a simple sine wave that starts at -15 db, increases to -5, then -4, back to -5, and all the way out to -15. (Warning: sine waves can be very shrill, so lower your volume before clicking on the following link)
Now let’s setup a compressor with a threshold of -10 db and a ratio of 5:1. The result should roughly be an output of -9 where -5 was, and -8.2 where -4 was. The -15 will remain the same because it is not reaching the compressor’s threshold. Only signals above the threshold get taken down.
You probably didn’t hear much of a difference did you? That’s because less than a few decibels of difference is barely going to be audible, even to practiced listeners. This does raise another good question though, which is why compress at all? Does it really do anything? Oh yes it does. In extreme forms it can be absolutely mad.
Let me prove this with a way more fun example. Here’s a few seconds of a track that I’ve been working on. I’ve placed on the entire track a compressor with a threshold of -3.5 db and a ratio of 3.5:1 (no particular reason, that’s just what sounded good at the time).
Song Sample at 3.5 ratio
I know there’s no harmony, cut me some slack. This was maybe two hours of straight work in FL.
The input signal, on average, comes in at around -8 to -5 db before the huge synth sounds happen. The kick transients sail straight up to -2 db, and the huge part is hovering between -1 to -0.1 db. This means that the only signals being compressed are the kicks and the drop, preserving some dynamic range. What’s dynamic range? Generally speaking its decibel difference between the highest and lowest amplitudes of a song.
With my song I’ve created a difference of about eight decibels. Let’s get rid of that with a limiter. A limiter is a type of compressor that has a ratio set so absurdly high that increases above threshold through it – they get completely flattened out. I’m going to really express the effect of compression by using the FL Limiter, which has a ratio of infinity:1. To those of you who are perplexed on how any sound could possibly exceed infinity, my response is: ‘exactly.’ It sounds impossible, but that’s not what it’s saying. The high ratio that defines the limiting function is telling us that, no matter how high the signal rises above threshold, the output level will always remain the same. An increase of 2 dB above threshold produces the same output as an increase of 20 dB. Or 200 dB, or (theoretically) an infinite amount of increase. In practice, an incredibly high input would overload the compressor’s input and become distorted. Or melt the thing into slag, insofar as ∞ dB represents the signal produced by a lightning bolt or something. In realistic terms, any ratio of 10:1 or above is limiting, because the change at output for signals over threshold becomes less and less perceptible as the ratio gets higher. I’m going to set the threshold to -7, about the bottom of the dynamic range, and listen in terror as my track gets flat-lined.
For the first few bars it sounds okay since the synth and percussion isn’t being touched all that much. But then…oh no. Why’d my synth line disappear under the kick?? Why does my kick sound like someone put a wet paper towel on it?? Most of all, where’s the build to the drop! Oh my god I’ve killed my track and it was only two hours old. In fact, we can even compare the two waveforms visually to see that…
…I have definitely squashed my track. Man that second one is ugly.
Let me clear up a bit what is happening. The kick, which meets -2 db at low frequencies, has to be pulled down to meet the -7 threshold which everything is being limited to. When that happens in a complex waveform with multiple frequencies combining at different amplitude levels, the result is that all the frequencies get pulled down the same amount. Compressors don’t discriminate by frequency. If the signal is going to be compressed by -8 db, then a 500 hertz saw wave already at -8 db is going to go down just as much as a 40 hertz bass kick at -2 db, so long as these two are occurring within the same auditory moment of the song. Since the kick I have reaches frequencies across the spectrum, the entire sound comes across muted, along with the synthesizer momentarily being ducking under the kick.
This is what people describe as over-compression. It’s when the compression works in such a way to create volume disparities across your mix. Who knew that having your track at a consistent volume level could make it sound briefly quieter?
So why would we ever use such heavy compression? Isn’t dynamic range something we want to preserve? Well yeah for sure, in an entire track. However, for individual instruments it’s not so bad an idea. Maybe you want a specific sound to be consistent in volume throughout a mix, such as a guitar or a singer (especially since singers need to stand out). In that case a limiter is a great way to keep a certain sound steady in the mix. As for the resultant lowering of volume caused by the limiter, you can make up for that with a function known as make up gain. It’s also sometimes called output control, but it is most colloquially known as raising the volume. Almost all compressors include a make up gain control with a lot of boosting capability for when you really compress a sound.
It’s not so bad either to use heavy compression on a track when there are only a few moments you want to have stand out, such as the kick drum. Even with the limiter on my track, it’s not impossible to get back some dynamic range on those kicks. Remember those four functions of compressors I talked about earlier? I’ve only mentioned two so far. The last two are attack and release. Attack is how long the compressor takes to begin responding to signals after they’ve reached the threshold, and the release is how long the compressor continues working after the signal passes beneath the compressor. Both of these are measured in milliseconds and allow you to time your compression’s output in such a way that pivotal impact sounds like kicks come through uncompressed while sustained instrumental sounds are limited. To test this out, here’s another version of my track with a heavy limiter of 20:1 (not as bad but still rather hard), an attack time of 10 ms.
Song Sample with Attack Controls
(also I added some stuff I’m sorry I couldn’t help it)
And another screenshot:
Still a bit dented, but the situation is way better than the heavy limiting was. The kicks and synths are better preserved for sure. That’s because the limiter is now providing room for the transients, the brief and sudden pumps in amplitude that represent the kick.
Now that you have a more complete understanding of how compressors can affect a sound, I’d suggest trying it out yourself. Try squashing the heck out of your tracks to hear what you risk losing in your mix. Try using the limiter as a volume control for individual sounds to understand exactly how the sound is altered. Soon enough you’ll be able to tell when compression is good for a track or not.