Module 4.3 · Tools Track · Compression
Compression Per Instrument
Vocals need slow attacks; drums need fast attacks; bass needs consistency. Each source has its own compression sweet spot — here are the recipes you'll reach for hundreds of times before instinct takes over.
You've learned the basics (Module 4.1 — what threshold, ratio, attack, release actually do) and the advanced techniques (Module 4.2 — parallel and sidechain). What you don't yet have is the muscle memory of which settings to start with for each kind of source. That's this module's job.
Vocals don't compress like drums. Drums don't compress like bass. Acoustic instruments don't compress like synths. The difference isn't taste — it's the physics of each source's transients and sustains. A snare drum has a 1 ms attack and 50 ms of body; a slow attack on a snare lets the transient through, then catches the body. A vocal has a 30+ ms attack and seconds of sustain; the same slow attack on a vocal misses the consonants entirely and only catches the held notes. Same control, totally different result.
The recipes below are starting points. They're the moves working engineers reach for when they hear "this needs compression." Adjust by ear, A/B against bypass, and let the source tell you what it needs.
⛓ Where this sits in the signal chain
Per-instrument compression sits at the standard compression stage in your channel chain: after subtractive EQ, before additive EQ. The cleanup cuts in Module 3.2 happen first; then this compression shapes dynamics and feel; then the additive moves in Module 3.3 add character to the now-controlled signal. For drums and vocals especially, members often stack a fast compressor (catches peaks) and a slow compressor (smooths sustain) — both at this stage of the chain.
First, the words
Four ideas about how compression "feels" different on different sources. Get these in your bones and the recipes below stop being rules and become instincts.
Concept 1
Compression "feel"
The same compressor with the same gain reduction can feel aggressive, gentle, punchy, squashed, or invisible — depending on how attack and release are set.
Think of it like steering a car: same wheel, totally different feels at different speeds.
A compressor isn't one tool — it's a tool whose character is shaped almost entirely by the attack and release knobs. Set the same compressor with 4:1 ratio and 6 dB gain reduction, then change attack from 1 ms to 30 ms, and the result transforms from "smashed and aggressive" to "punchy and breathing." Change release from 50 ms to 500 ms and the result transforms from "fast pumping" to "smooth and invisible." The dB of gain reduction tells you HOW MUCH you're compressing; the attack and release tell you HOW IT FEELS. Most compression learning isn't about ratio — it's about developing an ear for what fast attack vs. slow attack does to a transient, and what fast release vs. slow release does to the sustain. The visual below maps the four "feel quadrants."
Concept 2
Transient vs. sustain
Every sound has a brief peak (the transient) and the longer body (the sustain) — compression treats them differently depending on attack speed.
Think of it like a hand clap (the transient) followed by the lingering air (the sustain).
A snare hit, a kick drum, a vocal consonant, a guitar pluck — every percussive sound has the same two-part shape: a sudden peak (the transient) followed by a smoother body that rings out (the sustain). The transient is what makes the sound "snap" or "punch"; the sustain is what gives it "body" or "warmth." A fast attack catches the transient — pulls down the snap, so the sound feels softer and less punchy. A slow attack lets the transient through — the snap stays full strength, but the sustain gets compressed afterward, which actually makes the transient feel more punchy by contrast. Drum compression is almost always slow-attack for this reason. Vocal compression is usually slow-attack for the same — let the consonants through, control the held vowels.
Concept 3
Release determines musicality
Release timing controls how the compressor "breathes" between hits — and the right timing depends on the song's tempo and the source's natural rhythm.
Think of it like the rhythm of someone's breath after running.
After a kick hits and the compressor pulls down, how fast does it return to full level? That's release time. Too fast and the compressor "pumps" audibly — clicks and breaths between hits. Too slow and the compressor stays clamped down, never releasing before the next hit, and the sustain gets choked. The sweet spot depends on the song. For a 120 BPM track with a 4-on-the-floor kick, each kick is 500 ms apart — release timed to about 200–300 ms lets the compressor recover cleanly between kicks. For a slow ballad with sustained notes, releases of 400–800 ms feel more musical. Release is the most context-dependent control; an "auto" release can be a good starting point but ear-tuning beats automation every time.
Concept 4
The two-compressor stack
For demanding sources (vocals especially), engineers stack two compressors in series: one fast for control, one slow for glue.
Think of it like a goalkeeper plus a sweeper — different jobs in the same play.
A single compressor can either control fast peaks OR smooth out long-term level — but it's hard to do both well at once. The pro move is to use TWO compressors in series. The first ("fast" or "leveler") catches sudden peaks with fast attack, fast release, and ratio around 4:1, doing 2–4 dB of gain reduction on transient hits. The second ("slow" or "glue") smooths long-term dynamics with slow attack, medium release, and gentle ratio around 2:1, doing 2–3 dB of gain reduction on sustained passages. The lead vocal benefits most from this stack — fast comp catches loud consonants, slow comp evens out the verse-vs-chorus level changes. The classic CLA setup: an 1176-style fast comp followed by an LA-2A-style slow comp. Stock plugin equivalents work just as well.
The visual below maps the four quadrants of compression "feel" — what happens when you combine fast/slow attack with fast/slow release. Each quadrant has its own character and its own use cases. This is the mental map every working engineer carries when they reach for a compressor.
The four quadrants of compression character. Most working engineers think this way unconsciously: "this needs punch" → top-right. "this needs glue" → bottom-right. "this needs body" → bottom-left. "this needs control" → top-left. The recipes below land in one of these quadrants for each source.
Per-instrument recipes
Working starting points by source. Always A/B against bypass. Always listen in mix context.
Lead vocal — fast + slow stack
| Stage | Ratio | Attack | Release | GR target | Style note |
|---|---|---|---|---|---|
| Fast comp (1st) | 4:1 | 3–10 ms | 50–100 ms | 3–6 dB on peaks | 1176-style — catches loud consonants |
| Slow comp (2nd) | 2:1 | 30 ms | 200 ms | 2–4 dB sustained | LA-2A-style — evens long-term level |
| Or single comp | 3:1 | 10 ms | 100 ms | 4–6 dB | simpler workflow if not stacking |
Backing vocals (bus)
| Stage | Ratio | Attack | Release | GR target | Style note |
|---|---|---|---|---|---|
| Bus comp | 2:1–3:1 | 10 ms | 150 ms | 3–5 dB | glue all backings together as one unit |
| Optional parallel | 8:1 | fast | fast | blend ~20% | adds cohesion + fullness |
Kick drum
| Goal | Ratio | Attack | Release | GR target | Style note |
|---|---|---|---|---|---|
| Body shaping | 4:1 | 20–30 ms | 80–150 ms | 3–6 dB on peaks | slow attack lets the click through, controls the body |
| Heavy parallel | 10:1 | fast | fast | 10+ dB on the parallel | blend 20–30% with the dry kick |
| Sub control | 3:1 | 10 ms | 100 ms | 2–4 dB | only on the sub frequencies (multi-band) if needed |
Snare drum
| Goal | Ratio | Attack | Release | GR target | Style note |
|---|---|---|---|---|---|
| Punch enhancement | 4:1 | 30 ms | 80 ms | 4–6 dB on hits | classic snare comp — the punch enhanced quadrant |
| Heavy parallel | 10:1 | fast | fast | 10+ dB | blend 25–35% with dry — adds body without losing snap |
Toms
| Goal | Ratio | Attack | Release | GR target | Style note |
|---|---|---|---|---|---|
| Tom shaping | 4:1 | 20 ms | 200 ms | 4–8 dB on hits | controls the long tail of the tom resonance |
| Gate before comp | n/a | n/a | n/a | n/a | tom mics get bleed; gate first, then compress |
Drum overheads
| Goal | Ratio | Attack | Release | GR target | Style note |
|---|---|---|---|---|---|
| Overheads | 3:1 | 30 ms | 150 ms | 2–4 dB | gentle — preserve cymbal detail |
| Room mics | 4:1+ | fast | slow | 6–10 dB | "squashed body" — bring up room sound |
Drum bus (group glue)
| Goal | Ratio | Attack | Release | GR target | Style note |
|---|---|---|---|---|---|
| Glue (SSL-style) | 4:1 | 10 ms | auto/fast | 2–3 dB | classic SSL G-Series setting; gentle, invisible |
| Heavier punch | 4:1 | 30 ms | 100 ms | 3–5 dB | more audible glue — kit feels more "together" |
Bass (DI or amp)
| Goal | Ratio | Attack | Release | GR target | Style note |
|---|---|---|---|---|---|
| Consistency | 4:1 | 10 ms | 80 ms | 4–8 dB | bass needs consistency — every note same level |
| Two-stage stack | 3:1 + 2:1 | fast + slow | various | 3 dB + 2 dB | fast catches plucks; slow evens sustain |
| Sidechain to kick | 4:1 | fast | ~150 ms | 3–6 dB | kick triggers — Module 4.2 |
Acoustic guitar
| Goal | Ratio | Attack | Release | GR target | Style note |
|---|---|---|---|---|---|
| Tame strums | 3:1 | 30 ms | 200 ms | 3–5 dB on peaks | controls dynamics without losing pick attack |
| Optional parallel | 4:1 | 10 ms | fast | blend ~20% | adds body for fingerpicked acoustic |
Electric guitar (clean)
| Goal | Ratio | Attack | Release | GR target | Style note |
|---|---|---|---|---|---|
| Even out picking | 3:1 | 10 ms | 100 ms | 2–4 dB | tames varying pick dynamics |
| Often skipped | — | — | — | 0 dB | clean guitar often doesn't need much compression |
Electric guitar (distorted)
| Goal | Ratio | Attack | Release | GR target | Style note |
|---|---|---|---|---|---|
| Often skipped | — | — | — | 0 dB | distortion is its own compressor — usually doesn't need more |
| If used | 3:1 | 10 ms | fast | 1–3 dB | only for taming inconsistent strums |
Lead guitar / solo
| Goal | Ratio | Attack | Release | GR target | Style note |
|---|---|---|---|---|---|
| Smooth + sustain | 4:1 | 10 ms | 200 ms | 4–8 dB | helps lead notes sustain and sit consistent in the mix |
Piano / acoustic keys
| Goal | Ratio | Attack | Release | GR target | Style note |
|---|---|---|---|---|---|
| Tame dynamics | 3:1 | 20 ms | 200 ms | 3–5 dB | piano has wide dynamic range; gentle comp helps it sit |
| Parallel for body | 4:1 | 10 ms | fast | blend ~15% | adds sustain to staccato notes |
Synth pads / leads
| Goal | Ratio | Attack | Release | GR target | Style note |
|---|---|---|---|---|---|
| Pads (often skipped) | — | — | — | 0 dB | synths come pre-compressed via sustain — usually no extra needed |
| If used (lead) | 3:1 | 10 ms | 100 ms | 2–4 dB | gentle leveling for lead synths only |
| Pad sidechain pump | 8:1+ | fast | tempo-sync | 6–10 dB | EDM pumping effect — Module 4.2 |
Mix bus / 2-bus
| Goal | Ratio | Attack | Release | GR target | Style note |
|---|---|---|---|---|---|
| SSL-style glue | 2:1 | 10 ms or 30 ms | auto | 1–3 dB on choruses | the classic mix-bus glue compression |
| Heavier mix bus | 4:1 | 30 ms | 100 ms | 3–4 dB | more obvious cohesion — used on dense mixes |
"Compression isn't 'more or less.' It's 'fast or slow,' 'aggressive or invisible.' The dB of gain reduction is just the size of the move; the attack and release are what the move feels like." — FTM, on the real lever in compression
How to listen for over-compression
The most common compression mistake is going too far — compressing until the source sounds "wrong" without realizing what wrong sounds like. Train your ear by listening for these specific symptoms:
- "Pumping" — you can hear the compressor breathing between hits. Release is too fast for the program material. Slow the release.
- "Choking" — the source feels squashed, like it's being held down. Too much GR (8+ dB) for the source. Pull threshold up so GR is in the 3–6 dB range.
- "Loss of impact" — drums or vocals feel less punchy with the comp on than off. Attack is too fast — catching the transients you wanted to keep. Slow the attack.
- "Lifeless" — the track has no energy variation. Too much compression at every dynamic level. Lighten up — compression is supposed to control extremes, not flatten everything.
- "Muddy" — sustain comes up so loud the mid-range gets congested. Compressor is bringing up unwanted bleed/noise. Use a tighter ratio or less GR.
The check: bypass the compressor. Does the track sound clearly worse? If yes, the compressor is helping. If no, you're either over-compressing or compressing the wrong thing. Trust your bypass A/B.
The two-compressor stack — when and how
Most home-studio engineers don't use two compressors in series; they should. The principle: each compressor does one job well, and stacking lets each be optimized for its job. The classic CLA setup:
- First compressor — fast, peak control. Settings: 1176-style. Ratio 4:1, attack 1–3 ms, release 50 ms, GR 2–4 dB on peaks. Job: catch loud consonants, drum-stick clicks, momentary spikes.
- Second compressor — slow, level smoothing. Settings: LA-2A-style. Ratio 2:1, attack 30 ms, release 200 ms, GR 2–3 dB sustained. Job: even out the verse-vs-chorus level, the loud-vs-quiet sections.
The result: total GR is small at any given moment (2–4 dB from one comp + 2–3 dB from the other = 4–7 dB total) but the SHAPING is more sophisticated than any single compressor could deliver. You get fast peak control AND slow level glue, both at light settings.
Stock plugin equivalents:
- Logic: Compressor (set to "FET" or "VCA" mode with fast settings) → Compressor (set to "Vintage Opto" with slow settings)
- Ableton: Compressor with "Peak" detection (fast) → Compressor with "RMS" detection (slow)
- Pro Tools: Avid Channel Strip's compressor → Dyn3 Compressor on slow settings
- Reaper: ReaComp (fast) → ReaComp (slow), or ReaXComp for multi-stage
- FL Studio: Fruity Limiter's compressor (fast) → Fruity Compressor (slow)
- Studio One: Compressor (fast) → Compressor (slow)
Common per-instrument compression mistakes
- Fast attack on drums. Squashes the snap you wanted. Always start with slow attack on percussive sources — 20–30 ms minimum.
- Slow attack on a limiter. The whole point of a limiter is fast attack to catch peaks. 0.1–1 ms attacks for limiting.
- Fast release on a vocal. Causes audible pumping on held notes. 100+ ms release is the floor for vocals.
- Aggressive GR on a synth pad. Synths are already compressed by their envelope; piling on extra compression flattens them entirely. Pads usually don't need any comp.
- Compressing the kit AND the bus. If every drum is heavily compressed AND the drum bus is heavily compressed, you've stacked 6+ compressors. Pull back on either the individuals or the bus.
- Forgetting makeup gain. Compression reduces level; if you don't compensate with makeup gain, the track gets quieter even though it sounds "bigger."
- Same settings every track. A "default" 4:1 / 10 ms / 100 ms preset doesn't work everywhere. Match the settings to the source.
- Compressing without listening to the mix. Compression often sounds great in solo and bad in mix, or vice versa. Always check both.