Compression is the tool I reach for on almost every channel, and it’s also the tool I see people misuse the most. Not because it’s hard — because most explanations of it are vague. “It glues the mix together” tells you nothing about where to set the attack knob.
So here’s the practical version. What a compressor actually does, what each knob means, starting settings for vocals, drums, bass and 808s, and the three techniques (sidechain, parallel, multiband) that cover 90% of real-world use.
Quick note: you don’t need expensive hardware for any of this. Your stock DAW compressor is fine, and if you make sounds in Serum 2, there’s a solid compressor right on its FX buses — I cover that in my Serum 2 effects guide.
What compression actually does
A compressor turns down the loud parts of a signal automatically. That’s it. The distance between the quietest and loudest moments of a track is its dynamic range, and a compressor shrinks that range so the level stays more consistent.
Two controls define the behavior:
- Threshold — the level where the compressor starts working. Anything below it passes through untouched.
- Ratio — how hard it turns things down once the signal crosses the threshold. At 2:1, a signal that goes 4 dB over the threshold comes out only 2 dB over. At 4:1, the same 4 dB overshoot comes out as 1 dB. Above 10:1 you’re basically limiting.
One habit that will save you endless frustration: watch the gain reduction meter, not the knobs. For gentle leveling you want 2–4 dB of reduction on the loudest hits. If the meter is pinned at 10 dB all the time, your threshold is too low and the track will sound squashed.
And always match the loudness with makeup gain when you A/B. Compressed-plus-louder always sounds “better” — that’s your ears lying to you, not the compressor helping.
Attack and release: where the sound actually changes
Threshold and ratio decide how much compression happens. Attack and release decide what it sounds like. This is where the tone shaping lives.
Attack is how fast the compressor clamps down after the signal crosses the threshold.
- Slow attack (10–100 ms) lets the initial transient through before the compression grabs the body of the sound. Result: more punch, more “in your face”. Great for snares, plucked instruments, anything where the click matters.
- Fast attack (0–5 ms) catches the transient too. The sound gets rounder, smaller, sits further back. Useful for pads, backing vocals, background elements — and for taming sharp clicks and harsh peaks.
Release is how fast the compressor lets go after the signal drops back under the threshold. Typical range is 40 ms to a few seconds.
- Fast release makes the compressor breathe with the material. It brings up sustain and room, adds energy, but push it too far and you hear pumping (which can be the point in club music).
- Slow release keeps gain reduction smooth and invisible. Better for vocals and buses where you don’t want to hear the compressor at all.
A trick that works on rhythmic material: set the release so the gain reduction meter returns to zero right before the next hit. The compressor then locks into the groove instead of fighting it.
Compressor types in plain terms
Every plugin folder is full of emulations named after circuit types. Here’s what the labels actually mean for the sound:
- VCA (like the SSL bus compressor or dbx 160) — fast, clean, precise. The all-rounder. My default for drum buses and mix bus glue.
- Optical (like the LA-2A) — slow and smooth by design, because the gain control literally depends on a light element. Forgiving on vocals, bass and guitars.
- FET (like the 1176) — extremely fast and aggressive, adds a bit of grit. Classic on snares, room mics and in-your-face vocals.
- Vari-mu (like the Fairchild) — tube-based, gentle, more of a “thickener” than a peak catcher. Common on mix buses and mastering chains.
Honest advice: at the start, the type matters far less than your attack and release settings. A stock digital compressor with the right settings beats a vintage emulation with the wrong ones every time.
Starting settings by instrument
These aren’t rules, they’re starting points. Set them, then adjust while looking at the gain reduction meter.
| Source | Ratio | Attack | Release | Gain reduction |
|---|---|---|---|---|
| Lead vocal | 3:1 | 10–30 ms | 100–300 ms | 3–6 dB |
| Kick / snare | 4:1 | 20–40 ms | 50–150 ms | 3–5 dB |
| Drum bus | 2:1–4:1 | 10–30 ms | fast / auto | 2–4 dB |
| Bass / 808 | 4:1 | 10–30 ms | 100–200 ms | 3–6 dB |
| Pads / keys | 2:1 | 1–5 ms | 200–500 ms | 2–3 dB |
| Mix bus | 1.5:1–2:1 | 30 ms | auto / 100–300 ms | 1–2 dB |
On vocals, one compressor doing 8 dB of work sounds obvious. Two compressors in series doing 3–4 dB each sound natural. That’s the oldest trick in vocal mixing and it still works.
For 808s specifically, compression is only half the story — saturation and a clean single-oscillator patch matter more. I wrote a full walkthrough on that in my Serum 2 808 bass tutorial, and the Synopsis 808 presets are already processed to sit at the front of a mix. If you’d rather work with samples, a well-recorded one-shot needs almost no compression at all — that’s the whole idea behind our 808 one-shot kit.
Sidechain, parallel and multiband compression
Sidechain compression means the compressor listens to a different signal than the one it’s processing. The classic move: put a compressor on the bass, feed the kick into its sidechain input, and the bass ducks every time the kick hits. That’s the pumping you hear in house, EDM and a lot of hip-hop. Start with a fast attack, a release around 100–250 ms (tune it to the tempo), and enough threshold for 4–8 dB of ducking. If you just want the kick and bass to coexist without an audible pump, 2–3 dB is plenty.
Parallel compression (New York compression) blends a heavily crushed copy of a track with the untouched original. Send your drums to a bus, compress that bus hard — fast attack, fast release, 10+ dB of reduction — and then bring the fader up under the dry drums until they feel dense but still punchy. You get sustain and weight without losing transients, because the original track still carries them.
Multiband compression splits the signal into frequency bands and compresses each one separately. It’s the right tool when only part of the spectrum has a dynamics problem: a bass whose low end wobbles from note to note, or a vocal with harsh 5–8 kHz peaks (that specific case is called de-essing). Just compress the problem band and leave the rest alone. If the whole track needs work, use a normal compressor — multiband everywhere is a fast route to a lifeless mix.
Common mistakes I still hear all the time
- Compressing to fix a level problem. If a part is just too loud in the chorus, that’s a fader or automation job. Compression is for moment-to-moment dynamics, not song-section balance.
- Fast attack on everything. Kill every transient and the mix turns to mush. When drums lose punch after compression, the attack is almost always too fast.
- Judging with the compressed version louder. Match levels before you decide anything.
- Stacking compressors on already-compressed sounds. Most modern sample packs and presets are processed to sit in a mix out of the box. Loops from something like Drums Utopia don’t need another 6 dB of squashing — a touch of bus glue is enough.
- No gain staging. If the signal hits the compressor way too hot, every threshold setting behaves differently than the preset designer intended. Aim around -18 dBFS average going in.
FAQ: audio compression
What do the attack and release knobs actually do?
Attack sets how quickly the compressor turns the volume down after the signal crosses the threshold; release sets how quickly it recovers afterwards. Slow attack keeps punch, fast attack rounds the sound off. Fast release adds energy and possible pumping, slow release stays transparent.
How much gain reduction should I aim for?
For natural results, 2–6 dB on the loudest moments of individual tracks and 1–2 dB on the mix bus. More than that isn’t wrong — parallel drum crushing lives past 10 dB — but past 6 dB on a lead vocal or bus, you’ll usually hear the compressor working.
Should I compress every track?
No. Compress when something has a dynamics problem: a vocal that jumps around, a bass with uneven notes, drums that lack density. A steady synth pad or a pre-processed loop often needs nothing. An untouched track always sounds more alive than a needlessly compressed one.
What’s the difference between a compressor and a limiter?
A limiter is a compressor with a very high ratio (10:1 and up) and a near-instant attack. It’s a ceiling — it stops peaks from passing a set level, usually at the very end of the master chain. A compressor shapes dynamics; a limiter caps them.
Where to go from here
Pick one song you’re working on right now and practice one technique per session: gentle vocal compression today, sidechain pump tomorrow, parallel drums after that. Watching the gain reduction meter while you turn knobs teaches you more in an evening than a week of articles.
And if you want raw material to practice on, grab the 250+ free Serum 2 presets — basses, keys, pads and 808s you can compress, sidechain and crush without spending anything. Run them through the settings from this post and you’ll hear exactly what each knob does.
I run Monosounds.studio and make the packs myself. Ten-plus years of producing, 97+ releases, 16,000+ downloads. Every Serum 2 preset we ship passes a CPU check on a mid-range laptop and gets used in a real track before release. If it does not survive a mix, it does not go in the pack.
More about Maxim & the studio →
