Compression: threshold, ratio, attack and release, properly explained
The four controls that decide everything, what each one does to the sound, and how to set them by ear instead of copying presets.
What compressing is, without metaphors
A compressor turns the signal down when it goes past a level you choose, by a proportion you choose. Nothing else. Everything else — that it «punches», that it «adds body», that it «glues the drums» — is a consequence of that single operation applied at different speeds. Hold on to the mechanics and the four controls stop being a mystery.
The underlying reason it exists: music has an enormous range between its softest and loudest moments, and that range doesn't fit through a car stereo, a phone or a bar. Compressing brings the extremes closer so the quiet parts are audible without the loud ones becoming a problem. Done well, the listener doesn't perceive compression: they perceive that they follow everything without touching the volume.
The four controls
| Control | What it decides | What happens if you overdo it |
|---|---|---|
| Threshold | The level at which it starts acting | Compresses constantly, including the part that didn't need it |
| Ratio | How much it turns down per dB above the threshold | Flattens the dynamics: everything at one level, no relief |
| Attack | How long it takes to react after the threshold is crossed | Fast kills the initial hit; slow lets peaks escape |
| Release | How long it takes to let go when the signal drops | Fast causes audible pumping; slow leaves the signal held down and dull |
Attack: where punch is won or lost
This is the control most people set backwards. Intuition says that if you want to control peaks you should react fast, and that's precisely what takes the force out of drums. A snare hit lasts a few milliseconds: if the compressor arrives during it, it takes out exactly the part that makes it sound like a snare and leaves the body, which is the dull part.
That's why a 10 to 30 ms attack on a snare usually sounds punchier than a 1 ms one, even with the same reading on the meter. The transient passes whole, the compressor arrives afterwards and holds the tail: the result is a hit that stands out more against a more controlled background. And the reverse: if you're taming an uneven bass, a fast attack is right, because there's no transient worth preserving.
Release and pumping
Release decides how fast the compressor gives the level back when the signal drops. Too short and it moves up and down fast enough that you hear the background of the mix breathing — that's pumping. Too long and it never lets go before the next hit, leaving the track held down and lifeless.
The most reliable starting point is the tempo: have the compressor let go just before the next hit. Set release by listening to the background, not the meter, and raise it until the movement stops drawing attention. In genres where pumping is part of the sound — plenty of electronic music — you do the exact opposite and make it audible on purpose.
How to set it without presets
- Start with a low ratio (2:1 or 3:1) and bring the threshold down until peaks show 3 or 4 dB of reduction.
- Set attack by listening to the hit: shorten it until punch disappears, then back off until it returns.
- Set release by listening to the background between hits: lengthen it until it stops breathing.
- Make up the output gain until it matches the bypassed level, and only then compare.
- If at matched loudness you can't tell which you prefer, the compressor wasn't needed.
The loudness matching in the fourth point isn't a detail: it's what separates judging the processing from judging the level. If that sounds like something you should have sorted before compressing, it's covered in the gain staging guide. And to train your hand on moving dynamics, the Gain Rider drills exactly that.
Frequently asked questions
- How many dB of gain reduction is correct?
- There's no correct figure, there's a consequence. 2 to 4 dB on peaks is transparent control you don't hear working. 6 to 10 is an audible effect that changes the character. More than that is an aesthetic decision, not a fix. What matters is knowing which of the three you're in, and meaning it.
- What's the difference between a compressor and a limiter?
- It's a matter of degree, not of kind: a limiter is a compressor with a very high ratio (10:1 upwards, effectively infinite) and a very fast attack. The compressor shapes dynamics; the limiter sets a ceiling the signal doesn't cross. They're used for different things even though the mechanism is the same.
- Isn't a fast attack always better for control?
- No, and it's the most expensive misunderstanding. A fast attack catches the transient — the initial hit — and flattens it, so the snare loses crack and the guitar loses pick. A slower attack lets that hit through and compresses what follows, which is usually exactly what you want: keep the impact, control the tail.
- What is parallel compression and why is it everywhere?
- It means blending the uncompressed signal with a heavily compressed copy instead of replacing one with the other. You keep the original's transients and add the body and density of the crushed version. It's how you gain weight without losing punch, which is why it's standard on drums and vocals.
- When do I compress and when do I automate the volume?
- Automation corrects level differences between sections or phrases; the compressor corrects differences between syllables and hits, which move too fast for a hand. The usual order is a rough automation pass before compressing: a compressor works far better when it isn't also fixing the long-range imbalances.
- Is there any point in two compressors in series?
- Yes, and it often sounds better than one working twice as hard. Two gentle stages — one catching peaks with a slow attack, one levelling the body with a low ratio — split the job and draw less attention than a single compressor pulling 8 dB. Same principle as several small corrections instead of one big one.