Audio Compression Explained: How Compressors Work and When to Use Them

Audio compression explained in one sentence: a compressor automatically turns a signal down whenever it gets louder than a level you set, narrowing the distance between its loudest and quietest moments. That smaller dynamic range lets a vocal sit steadily in a mix, a kick drum hit harder and a full mix feel glued together. Used without listening, the same tool flattens the life out of a performance.

Audio compression explained: what does a compressor actually do?

A compressor is an automatic volume control. It measures the incoming level, and when that level crosses a point you choose, it reduces the gain by an amount you choose.

The result is reduced dynamic range, the difference between the quietest and loudest parts of a sound. If a singer whispers the verse and belts the chorus, a compressor brings the belted notes down so you can raise the whole vocal and still hear the whispers. (This is unrelated to data compression like MP3, which shrinks file size.)

Engineers use it for control (steady, audible levels), for tone and impact, and for glue on a drum bus or the mix bus.

Fab Dupont, award-winning producer, engineer and Campus mentor whose credits include David Crosby, André 3000 and Bad Bunny, names the classic mistake: "You know what's a terrible compression strategy? Scrolling through presets until one sounds less bad." His point: compression "actually makes perfect sense once you understand what you're hearing and most importantly what you're doing."

How do threshold and ratio work?

Threshold sets the level where compression begins, and ratio sets how strongly the compressor reduces anything above it. Together they determine the amount of gain reduction.

• Threshold: signal below it passes untouched. Lower it and more of the signal gets compressed.
• Ratio: at 4:1, for every 4 dB the signal rises above the threshold, only 1 dB comes out.
• Knee: a hard knee applies the full ratio the instant the threshold is crossed; a soft knee eases in and sounds gentler.
• Gain reduction meter: shows how many decibels the compressor is turning down at any moment.

A worked example: threshold at −20 dB, ratio at 4:1, and a vocal peak hits −12 dB. The peak is 8 dB over; at 4:1 it comes out 2 dB over, at −18 dB, so the gain reduction is 6 dB.

Low ratios (around 1.5:1 to 3:1) with a low threshold give gentle, overall leveling. High ratios (8:1 and up) with a higher threshold catch only the peaks.

What do compressor attack and release do?

Attack is how quickly the compressor reacts once the signal crosses the threshold; release is how quickly it lets go after the signal falls back. They decide whether compression sounds punchy, smooth, invisible or pumping.

Fab Dupont puts it plainly: "Attack and release aren't just for dog training, they're also the two most important compressor settings." Once you understand them, he says, "you can stop guessing and start truly shaping tone, punch and even groove on purpose instead of by accident."

Attack

• Fast attack (under 1 ms to a few ms) clamps down on transients: drums lose snap, vocals get smoother, bass gets rounder. Too fast sounds dull and flat.
• Slow attack (roughly 10 to 30 ms or more) lets the initial transient through first, which adds punch to drums, picked bass or acoustic guitar.

Release

• Fast release brings up quieter detail and room sound. Too fast causes pumping, or distortion on low frequencies.
• Slow release is smoother, but if the compressor never recovers between notes, the track stays squashed.
• Release timed to the song lets the meter fall back with the beat, so compression breathes with the groove instead of fighting it.

What is makeup gain, and why can it fool your ears?

Makeup gain (or output gain) raises the level after compression to compensate for the gain reduction. Without it, a compressed track simply sounds quieter.

The trap is that ears tend to prefer whatever is louder. Add 4 dB of makeup gain for 3 dB of gain reduction and the compressed version will seem better for the wrong reason. Set it so bypassing the compressor causes no jump in loudness, then compare. If you can't hear an improvement at equal level, the compressor isn't helping. Input level matters as much as output level; see what is gain staging.

What are the main types of compressors?

The four classic types of compressors are VCA, FET, optical and variable-mu. All reduce gain above a threshold; the circuit gives each its own speed and character.

Digital compressors, including your DAW's stock plugin, can be designed to behave like any of these. That's why Fab Dupont doesn't send anyone shopping: "Whether it's a vintage emulation or the stock compressor in your DAW, the principles are the same. Learn the fundamentals once and then every compressor suddenly becomes your friend." To see one of these designs up close, Fab Dupont explains the LA-2A's photocell in how optical compressors work on Puremix, and Process.Audio's COMP.TWO is a plug-in emulation of an all-tube vari-mu compressor.

How do you compress vocals?

Here's how to compress vocals: start with a moderate ratio, set the threshold for a few decibels of gain reduction on the loudest phrases, then shape attack and release by ear while the full mix plays. These numbers are a starting point, not a recipe. For the full chain around the compressor, see how to mix vocals.

1. Fix the big jumps first. Lower any word that's far louder than the rest with clip gain or automation.
2. Set a moderate ratio, around 3:1 to 4:1.
3. Lower the threshold until the loudest phrases show roughly 3 to 6 dB of gain reduction.
4. Adjust attack. If word starts lose clarity, slow it down; if syllables still poke out, speed it up.
5. Adjust release so the meter recovers between phrases instead of staying pinned.
6. Level-match and toggle bypass in the mix, not in solo. If the vocal sounds smaller, flatter or breathier, back off.
7. Consider two stages: a fast compressor catching peaks, then a slower, gentler one for overall leveling.

Limiter vs compressor: what is the difference?

A limiter is a compressor with a very high ratio (typically 10:1 and above, up to effectively infinite) and a very fast attack, built to keep the signal under a ceiling. A compressor shapes dynamics; a limiter sets a hard maximum.

Compressors handle tone, control and glue; limiters catch stray peaks and, on a master, raise loudness. Pushed hard, a limiter erodes transients, so level-match and compare. Andrew Scheps compares the two on bass, alongside parallel compression, in Limiter vs Compressor on Puremix. For mix bus and multi-bus compression, see The Brauerize Mixing Method.

When should you not use compression?

Knowing when not to use compression matters as much as any setting. Skip it when the dynamics are the point of the performance, when the real issue is level or EQ, or when the track already sits well.

• The performance breathes on its own: solo piano, acoustic jazz or a fragile ballad vocal may depend on wide dynamics.
• The problem is one spot: a single loud note is a job for automation.
• The problem is tone: compressing a harsh sound mostly gives you a consistently harsh sound. Reach for EQ.
• Bypass sounds the same or better at matched level: remove it.
• You hear side effects: unintended pumping, lost transients, exaggerated breaths, or a mix that feels tiring.

How do you train your ears to hear compression?

You train your ears by exaggerating settings until the effect is obvious, then backing off and comparing at matched level. Fab Dupont's starting question is simple: "can you actually hear what your compressor is doing?"

That ears-first approach is the backbone of Foundations of Compression, Fab's 10-hour, all-levels course on Campus @ Puremix. It covers how compressors work but also when, why and how much to use them, with decisions made in context rather than from presets or formulas, across two modules: The Building Blocks of Compression and Practical Compression Strategies. Here's a drill in the same spirit.

Practice exercise: hear attack and release in 15 minutes

1. Put your DAW's stock compressor on a drum loop or a dry vocal.
2. Set the ratio to 4:1 and lower the threshold until you see 8 to 10 dB of gain reduction. This is deliberately too much.
3. Sweep attack from fastest to slowest. When does the snap come back?
4. Set attack in the middle and sweep release. When does the tail pump, and when does it stay clamped?
5. Raise the threshold to 2 to 4 dB of gain reduction, level-match and toggle bypass. Write one word for what changed: punchier, smoother, darker?
6. Repeat on a bass part; the same settings will behave differently.

Frequently asked questions about audio compression

What are good compressor settings for beginners?

A neutral starting point for many sources is a 3:1 or 4:1 ratio, a medium attack around 10 ms, a medium release around 100 ms, and a threshold that gives 3 to 6 dB of gain reduction on the loudest parts. From there, adjust attack and release by ear in the context of the mix.

How much gain reduction is too much?

There is no fixed number. Gentle leveling often sits around 2 to 4 dB of gain reduction, while deliberate, aggressive effects on drums or room mics can go well past 10 dB. It's too much when you hear side effects you didn't intend: lost transients, pumping, exaggerated breaths or noise, or a part that sounds smaller than the bypassed version at matched level.

Should I compress before or after EQ?

Both orders are valid. EQ before compression changes what the compressor reacts to; cutting excess low end first stops bass energy from triggering heavy gain reduction. EQ after compression shapes tone without changing how the compressor behaves. A common approach is corrective EQ first, then compression, then tonal EQ. Try both and choose by ear.

What is parallel compression?

Parallel compression blends a heavily compressed copy of a signal with the uncompressed original, usually through an aux send. The compressed copy brings up body, sustain and detail, while the dry signal keeps the natural transients. It's widely used on drums, vocals and bass. Many plugins include a mix or dry/wet control that does the same thing inside a single instance.

To build these skills with structure and feedback, Foundations of Compression covers the controls, compressor types and real mixing decisions with Fab Dupont, through video lessons, readings, listening exercises, quizzes and discussions. It's on-demand, with lifetime access, a teaching assistant, community forums, a certificate of completion and a 14-day money-back guarantee. Learner feedback is on the Campus reviews page.