A few articles back I wrote that among the thousand things an aspiring producer has to learn there's "the COMPRESSOR", all caps. That wasn't a typo. In my head, the compressor lived exactly like that for years: a black box you either learn to tame, or you keep wobbly dynamics for all your tracks.
Then one day I read the right explanation, the kind that makes you feel dumb for not getting it sooner: a compressor doesn't do anything you wouldn't already do with your mouse on the fader in Ableton. It just does it in milliseconds, with surgical consistency.
Mystery solved. Almost.
Because then come the parameters, and that's where you go back to racking your brain trying to make sense of any of it. After courses, dynamics gurus, and "a few" hours spent on Ableton, here are some notes on what I've figured out:
It's an automatic volume control, not an effect
Start here, because it's obvious, though not a given: a compressor reduces the dynamic range of a signal (for non-engineers: the distance between its quietest point and its loudest). It doesn't add or take away anything in terms of content: it raises the floor and lowers the ceiling. For example: a vocal that goes from a whisper to a scream in a short span of time sounds more even after compression, more "controlled".
Everything starts from a threshold: the level above which the compressor kicks in. Below that threshold, the signal passes untouched.
Above it, the ratio comes into play.
...
"Huh?"
Yeah, I know. But I promise it's less complicated than it looks.
The five parameters that actually matter
Threshold. The trigger point. The lower you set it, the more signal ends up inside the compression.
Ratio. Defines how many dB get shaved off whatever crosses the threshold, but expresses it as a fraction. In practice: with a 4:1 ratio, every 4dB of signal over threshold comes out as 1dB. At 1:1, nothing happens; toward infinity:1 the compressor stops compressing and starts limiting -- a hard cutoff, nothing gets past that point anymore.
Attack. How long it takes the compressor to kick in once the signal crosses the threshold. Fast, and you squash the note's attack, the thing that gives it "punch". Slow, and you let the initial hit through untouched, compressing only the tail. On a drum, two milliseconds of difference is the distance between a kick that slaps and a kick that just deflates.
Release. Defines how long the compressor takes to stop compressing once the signal drops back below threshold. Too short and you hear something like an unnatural breath; too long and the compressor stays active even when it shouldn't, squashing whatever comes right after.
Makeup gain. Obviously compressing has consequences -- it lowers the average volume, by definition. Makeup gain gives it back, bringing the compressed signal to the same perceived level as the original -- this is the parameter I use to check by ear whether I actually like the compressor, or whether I only like it because it's louder now.
Anything sounds "better" if it's louder.
And I don't want to fall for that.
Types exist, but you don't need to worship all of them
What personally threw me off a lot at the start is that once you start reading up on sound dynamics and compressors, you keep running into names like FET, VCA, Optical (Opto), or Vari-Mu/tube. And that did nothing but pile on more layers of information, more chapters I had to learn. They're not aesthetic whims: these "virtual" compressors, in plugin form, try to emulate real hardware compressors, each with different sonic behavior.
The FET type is very fast and aggressive.
The Optical type (like the LA-2A) is slow, "glue-like".
The practical truth is that we could talk about the history of compressors and what each one does, along with the plugin that emulates it. In reality, though, you don't need to own all of them -- the priority is understanding what the compressor is doing right now, on your track.
The color comes later, once you know which direction the track is taking and what you're actually looking for.
Where it's used (and what it's actually for)
On a single track, the compressor makes constant what a musician -- or me, recording at two in the morning -- can't play constantly on their own.
On buses and groups, it "glues" multiple tracks together, making them feel like one coherent thing instead of a collection of disconnected elements.
In sidechain (another fancy-sounding name), you use the volume of one track -- typically the kick, though that's not a fixed rule -- to automatically duck another one (bass, pads) every time the first one hits.
This last little trick is why certain electronic productions "breathe" in a way that seems impossible to program by hand, when really it's just a compressor listening to the kick for you.
The really nice part, though, is that in the end these are just some of the guidelines I've picked up over the years and have definitely used -- but nothing stops you from using them creatively, outside the "rules".
When to leave it off instead
It's important to understand when to compress, but it's just as important to know when not to. If a track already has natural dynamics that work -- a tom with a nice tail, a pad that rises and falls on its own -- compressing it often just flattens it, it doesn't improve it. I've done it plenty of times because "that's just what you do", but you need to stop before that: a compressor solves a specific problem (uncontrolled dynamics, transients slipping through, lack of cohesion), it's not a flavor to smear everywhere out of habit.
The lesson I'm taking home
The compressor that used to scare me wasn't that complicated, in the end. It was just badly explained, or if you prefer, explained in a way that never stuck with me -- tossed into a list of things "you should know" with nobody really telling me where to start.
All it took was starting with threshold and ratio, listening to what changed -- which is really the most important part -- and the rest followed on its own.
If you've got a word written in all caps somewhere in your head too, try rewriting it in lowercase. Usually, that's all it takes.