How Compression Works: Threshold, Ratio, Attack and Release

Compression reduces the difference between a sound's loud and quiet parts. It is one of the most important tools in mixing and recording because it controls dynamics, adds perceived punch, and helps vocals and bass sit steadily in a track. Compression can be subtle and nearly invisible, or heavy and obvious, and the difference depends on a few controls. This guide explains how it works and how to set those controls.

What a compressor does

A compressor watches the level of a signal. When the level rises above a set point called the threshold, the compressor turns the gain down by an amount set by the ratio. The loud parts come down, and the quiet parts stay the same. After the compressor, the whole signal can be raised with makeup gain to restore the overall level.

The result is a narrower dynamic range. The difference between the softest and loudest moments shrinks, so the sound feels more even and often more forceful. This is why compression is used on vocals, which change a lot in level from phrase to phrase, and on bass, which can jump in level between notes.

The threshold

The threshold sets the level at which compression starts. Signals below the threshold pass unchanged. Signals above it are reduced. A low threshold means more of the signal is compressed, and a high threshold means only the loudest peaks are affected. Set the threshold by playing the source and watching the gain reduction meter, which shows how much the compressor is working.

The ratio

The ratio sets how strongly the compressor reduces the signal above the threshold. A ratio of 2:1 means that for every 2 dB the input rises above the threshold, the output rises by only 1 dB. A ratio of 4:1 means the output rises by 1 dB for every 4 dB of input increase. Higher ratios reduce peaks more firmly, and very high ratios of 10:1 or more act like limiters.

A simple calculation shows the effect. Suppose the threshold is -20 dB, the ratio is 4:1, and the input is at -8 dB. The input sits 12 dB above the threshold. Divided by the ratio, the output sits 3 dB above the threshold, which is -17 dB. The compressor has reduced the signal by 9 dB at that moment.

Attack and release

Attack is how quickly the compressor responds once the signal crosses the threshold. A fast attack catches transients quickly and can soften a sharp drum hit, while a slow attack lets the initial transient through and then compresses the body of the sound. Release is how quickly the compressor lets go after the signal falls below the threshold. A fast release can make the sound pump or distort, while a slow release can leave the sound squashed for too long.

Starting points depend on the source. For a vocal, many engineers begin with a moderate attack and a release that recovers between phrases. For drums, a slower attack lets the snap through, and a release timed to the tempo keeps the groove. Listen closely, because the best settings often sound natural when they are almost inaudible.

The knee

The knee controls how the compressor transitions into compression around the threshold. A hard knee applies full compression as soon as the threshold is crossed, which is sharp and obvious. A soft knee eases into compression over a range of levels, which sounds smoother. Soft knees are common on mix buses and vocals, where a gentle curve is often wanted.

Makeup gain

Compression lowers the overall level, so makeup gain restores it. Without makeup gain, a compressed track sounds quieter than the original, and that difference makes it seem worse even when the compression is good. Compare the processed and unprocessed versions at matched loudness, because a louder sound usually sounds better regardless of quality.

Typical settings by source

The table below lists common starting points. Treat them as a place to begin, then adjust by listening.

Source	Ratio	Attack	Release	Purpose
Vocal	3:1 to 4:1	Medium	Medium	Even level and steady presence
Bass guitar	4:1 to 6:1	Medium to slow	Medium	Consistent low end
Drum bus	2:1 to 4:1	Slow	Auto or tempo-synced	Glue and punch
Acoustic guitar	2:1 to 3:1	Medium	Medium	Controlled dynamics
Mix bus	1.5:1 to 2:1	Slow	Auto	Gentle cohesion

Parallel compression

Parallel compression mixes a heavily compressed copy with the original, unprocessed signal. The compressed copy adds density and sustain, while the dry signal keeps the natural transients. Achieve it by sending a track to an auxiliary bus with a strong compressor, then blending that bus under the dry track. Parallel compression is a common technique on drums and vocals because it adds weight without flattening the performance.

Sidechain compression

Sidechain compression lets one signal control the compressor on another. The classic example is a kick drum triggering compression on a bass or pad, which makes room for the kick to hit clearly and produces a pumping rhythm. Most DAWs support sidechain routing through a plugin input. It is a powerful tool for dance music and for keeping vocals intelligible over busy backgrounds.

Compression is not a loudness fix

Compression can change perceived loudness, but it does not replace a proper level balance. Using compression to hide a weak performance or a poorly placed microphone usually makes problems more audible later. Fix the source first, then use compression to shape what is left. Automation, rides, and clip gain often solve level issues more transparently than a compressor.

A practical workflow

Set the makeup gain to zero. Choose a moderate ratio, such as 3:1 or 4:1, and set the attack and release near the middle of their ranges. Play the source and lower the threshold until the gain reduction meter shows a few decibels of action on the loud parts. Adjust the attack and release by ear until the sound feels controlled but natural. Finally, raise the makeup gain to match the original level, and bypass the compressor to compare.

Listening for over-compression

Over-compression shows up as a lifeless sound, a sense that everything is pushed to the same level, or pumping that follows the beat in an unwanted way. Signs of too much compression include a vocal that loses breath and room sound, drums that sound squashed, and a mix that feels tiring after a few minutes. A useful check is to bypass the compressor and then reinsert it, watching the gain reduction meter. Most useful compression stays within a few decibels of reduction on the peaks. Heavy reduction on every hit is usually a sign that the source needs a level fix, a better performance, or a different tool.

Key takeaways

Compression reduces the difference between loud and quiet parts. The threshold, ratio, attack, release, knee, and makeup gain control the effect, and each one changes how the sound feels.

Use compression to shape dynamics, not to fix problems that belong at the source. Parallel and sidechain techniques add weight and space, and matched-loudness comparisons keep your judgments honest.