A compressor is a dynamics processor that automatically reduces the level of a signal once it rises above a chosen threshold, by an amount set by the ratio. In audio mixing, it narrows the gap between the loudest and quietest moments of a source, which keeps vocals, bass and drums steady in level and lets them sit at a predictable place in the mix.
Compression is a form of automatic gain control with musical side effects. Beyond simple level control, engineers use a compressor to shape transients, add perceived density and glue several sources together, which is why it is one of the most frequently used tools on a mixing console after EQ.
How a compressor reduces dynamic range
A compressor has two parts: a detector (sidechain) that measures the incoming level and a gain element that turns the signal down. The detector compares the measured level with the threshold; whenever the level exceeds it, the gain computer calculates how much gain reduction to apply, and the gain element applies it in real time.
The ratio defines the static curve. With a 4:1 ratio, a signal that rises 8 dB above the threshold leaves the compressor only 2 dB above it, so 6 dB of gain reduction is applied. Because compression lowers the loudest parts, the overall output is quieter, and makeup gain restores the average level, which makes quiet details louder relative to the original.
Timing determines how the curve is applied to real music. Attack sets how quickly gain reduction begins after the threshold is crossed, and release sets how quickly the gain returns to unity once the signal falls. The knee controls whether compression starts abruptly at the threshold (hard knee) or blends in gradually over a range of several decibels (soft knee).
Key compressor parameters and typical settings
The following table lists the core controls of a compressor, their usual ranges in mixing and the audible effect each one has on the signal.
| Parameter | Typical range | What it affects |
|---|---|---|
| Threshold | -40 dBFS to 0 dBFS | Level where compression begins |
| Ratio | 1.5:1 to 10:1 | Strength of gain reduction above threshold |
| Attack | 0.1 ms to 100 ms | How much of each transient passes through |
| Release | 20 ms to 1 s or auto | How fast gain recovers; risk of pumping |
| Knee | 0 dB (hard) to about 12 dB (soft) | Smoothness of the onset of compression |
| Makeup gain | 0 dB to about 12 dB | Restores average level after reduction |
Most mixing tasks need only 3 to 6 dB of gain reduction on the loudest passages. Gain reduction meters show the result, but the decisive check is level-matched listening: when makeup gain makes the processed signal louder, the ear tends to prefer it simply because it is louder, not because it is better.
Compressor designs and their character
Analog compressors are grouped by the gain-reduction element they use, and each design has a recognizable response that digital emulations also reproduce. The table below compares the classic types.
| Design | Response | Typical use in mixing |
|---|---|---|
| VCA | Fast, precise, predictable | Drum bus, mix bus, punchy drums |
| FET | Very fast attack, aggressive color | Rock vocals, room mics, bass |
| Optical | Smooth, program-dependent release | Vocals, bass, gentle leveling |
| Variable-mu (tube) | Slow, soft, ratio rises with level | Mix bus glue, mastering-style smoothing |
| Digital look-ahead | Can react before the transient arrives | Transparent control, broadcast, dialogue |
Detection method also matters. Peak detection reacts to instantaneous level and catches fast transients, whereas RMS detection follows average energy and behaves closer to perceived loudness, which is why RMS-style compressors often sound more natural on sustained material such as vocals and pads.
Compressor versus limiter, gate and related processors
A compressor sits in a family of dynamics processors that differ in ratio, purpose and how they react to level. The comparison clarifies where the compressor ends and its neighbors in mixing begin.
| Processor | Ratio or action | Main purpose |
|---|---|---|
| Compressor | Moderate, about 1.5:1 to 10:1 | Reduce dynamic range and shape tone |
| Limiter | Very high, 20:1 to infinity:1 | Stop peaks from exceeding a ceiling |
| Noise gate | Attenuates signal below threshold | Remove bleed and noise in pauses |
| De-esser | Frequency-dependent compression | Tame sibilance around 4 to 10 kHz |
| Sidechain compression | Triggered by another signal | Duck one source when another plays |
The boundary between a compressor and a limiter is a matter of degree: above roughly 10:1 to 20:1, compression behaves like limiting. A noise gate works in the opposite direction, acting on signals below the threshold, and a de-esser is simply a compressor whose detector listens only to a narrow high-frequency band.
Setting a compressor in a mix
A reliable workflow is to set a high ratio and fast attack temporarily, lower the threshold until gain reduction is obvious, and then adjust attack and release by ear before returning the ratio to a musical value. Exaggerating the effect first makes the timing behavior audible, which is otherwise hard to judge at subtle settings.
Release should breathe with the tempo of the music: a release that recovers just before the next beat keeps the source lively, while one that is too short causes distortion on low frequencies. Many engineers also use parallel compression, blending a heavily compressed copy with the dry signal to add density while preserving the original transients.
| Source | Starting settings | Goal |
|---|---|---|
| Lead vocal | 3:1 to 4:1, attack 5 to 15 ms, release 50 to 150 ms | Even level, clear diction |
| Bass guitar | 4:1, attack 10 to 30 ms, release 100 to 200 ms | Consistent low end |
| Kick drum | 4:1, attack 20 to 40 ms, short release | More punch from preserved transient |
| Mix bus | 2:1, attack 10 to 30 ms, auto release | Gentle glue, 1 to 2 dB reduction |
Common compression mistakes
Most problems with a compressor come from too much gain reduction or poorly chosen timing. The list below covers the faults that appear most often in mixing.
- Attack too fast removes the transient and makes drums and plucked instruments sound dull and distant.
- Release too short follows individual waveform cycles and causes low-frequency distortion on bass and kick.
- Audible pumping occurs when a loud element repeatedly pulls down everything else on a bus.
- Raised noise floor results from makeup gain, which lifts breaths, hiss and room noise along with the quiet details.
- Loss of musical dynamics happens when every source is compressed hard, leaving a flat and fatiguing mix.
In simple terms
A compressor is like a very fast hand on a volume fader. When a singer suddenly gets loud, the hand pulls the volume down a little. When the singer gets quiet again, the hand lets the volume come back up.
The result is a performance that stays at a steadier level, so the listener does not have to strain to hear soft words or flinch at loud ones. Used gently, it is almost invisible; used heavily, it becomes an obvious effect that makes sounds feel dense and powerful.
A practical example
A mixing engineer is working on a pop song in which the lead vocal varies by about 12 dB between verses and choruses. Quiet lines disappear under the guitars, while loud chorus notes jump out of the mix.
The engineer inserts an optical-style compressor on the vocal with a 3:1 ratio, sets the threshold so the loudest notes receive about 5 dB of gain reduction, and chooses a 10 ms attack so consonants keep their definition. A second, faster FET-style compressor catches the occasional sharp peak with 2 dB of reduction.
After adding makeup gain and comparing at matched levels, the vocal sits consistently on top of the arrangement in every section. The engineer then needs far less fader automation, and the vocal remains intelligible on small speakers and in a car.