Frequency State / Music Production Reference Library · Volume 5
Compression controls the difference between stronger and weaker moments in a signal by changing gain dynamically. Used carefully, it can steady a vocal, change the impact of drums or help instruments coexist. The key is understanding the timing and detector behaviour, not choosing one universal setting for everything.
What a compressor does
A compressor is a dynamic processor that reduces the level of a signal when a specified condition is met. In most conventional downward compressors, that means the incoming signal crosses a threshold and the processor applies gain reduction. Its output can then be raised with makeup gain so the processed sound sits at a useful level in the mix.
Compression can make changes in level easier to control, alter the envelope of a drum hit or bring quieter details forward relative to louder passages. It cannot substitute for a balanced performance, careful automation or good recording technique. A vocal with wildly different phrases, for example, may benefit from clip-gain adjustments before compression.
Threshold and ratio
Threshold is the reference level at which compression begins to operate. Ratio describes how strongly the processor reduces levels above the threshold. With a 4:1 ratio in a simple hard-knee compressor, a steady input 8 dB above threshold would ideally emerge about 2 dB above threshold, giving roughly 6 dB of gain reduction before other effects of attack, release and detector behaviour.
The threshold is not a target recording level and the ratio does not describe the overall mix balance. A low threshold with a modest ratio can produce substantial compression over much of a performance; a high threshold and strong ratio may only catch occasional peaks. The gain-reduction meter is useful, but always compare the sound, not just the numbers.
Attack: what happens at the beginning
Attack determines how quickly the compressor responds after the detector identifies a signal above threshold. A relatively slow attack can allow an initial drum transient through before the gain reduction develops, helping preserve or emphasise impact. A faster attack may restrain that transient and produce a rounder or more controlled sound.
Attack controls are not perfectly standardised between compressors. One plugin may quote the time to reach a particular fraction of gain reduction, while another measures a different point. For that reason, the same millisecond figure does not guarantee the same audible result across plugins.
Release: what happens after the loud part
Release determines how quickly gain reduction relaxes after the detector level falls back. A very short release can create audible pumping or distortion on low-frequency material. A very long release can keep quieter sounds held down after a transient, causing the track to feel flat or sluggish.
Try listening to the gain-reduction meter alongside the rhythmic phrase. On a bassline, a release that follows the groove can sound controlled without obvious movement; on a sustained pad, a longer release may feel smoother. Some compressors offer automatic release settings that respond to programme material.
Knee, detector and sidechain
A hard knee begins compression more abruptly at the threshold, whereas a soft knee introduces the effect gradually around it. The latter may feel less obvious for vocals or complex instruments. Different peak or RMS-style detectors also react differently to transients and average signal energy.
The sidechain is the signal the detector listens to. Usually this is the track itself, but an external sidechain can make a bass instrument duck when the kick drum hits. A high-pass filter in the detector path can stop deep bass energy from dominating the compressor’s response without directly removing bass from the audible track.
Makeup gain and the loudness trap
After compression, the output may be quieter because the loudest passages have been reduced. Makeup gain restores the overall level, but a louder comparison almost always feels impressive at first. Match the processed and bypassed output levels as closely as you can before deciding whether the compression improved the sound.
Compression does not always make a mix more dynamic or punchier. Heavy gain reduction may remove useful contrast and increase the apparent prominence of noise or reverberation. When a processor is only being used to raise level, another gain control may be the cleaner tool.
Practical example: a vocal
Imagine a vocal with mostly consistent phrases and a few words that jump out. First correct any extreme individual words with clip gain or automation. Then try a moderate ratio, set threshold so the loudest phrases receive several decibels of reduction, and adjust attack and release so consonants remain intelligible and phrases recover naturally.
There is no universal vocal preset. A breathy performance, forceful spoken voice and energetic rock vocal need different responses. If the compressor exaggerates breath noise or makes the singer seem lifeless, reduce the amount and revisit the original level balance.
Practical example: kick and snare
On a drum hit, deliberately compare a quick attack with a slower one while keeping threshold and output level similar. The faster setting may soften the initial impact; the slower setting may let more of the transient through. Release changes how the hit returns to its original level before the next beat.
Parallel compression is another approach: blend a heavily compressed signal with an uncompressed or less-compressed version. This can add body while retaining some transient character, but latency differences and phase behaviour must be handled by the DAW or plugin routing.
Practical example: controlling the bass
Bass instruments can produce large energy variations between notes, particularly when different pitches or playing techniques interact with the instrument and room. Moderate compression can reduce the most distracting changes, but a bass sound with too much sub energy may require tonal correction as well.
A kick-triggered sidechain compressor is useful when kick and bass compete at the same moment. The aim need not be an obvious dance-music pumping effect: subtle ducking can simply create a brief opening for the kick.
When not to compress
If a part already has a stable envelope and sits clearly in the mix, compression may not help. Some sampled drums, heavily processed synth patches and sustained pads arrive with dynamics already shaped. Adding more processing can introduce distortion, noise emphasis or unnecessary loss of impact.
Try adjusting the performance, the instrument’s own amplitude envelope, clip gain or arrangement first. A simpler mix can sound clearer precisely because fewer processes are changing each signal.
A controlled listening experiment
Choose a short drum loop, vocal phrase or bassline. Listen to the original at a moderate level, add a compressor and begin with a ratio around 3:1. Lower the threshold until gain reduction becomes noticeable, then compare short and long attacks. Adjust release while observing whether the gain reduction returns between events.
Next set makeup gain so bypassed and processed versions sound roughly equally loud. Take notes about the attack, sustain, groove and apparent distance rather than judging only whether the effect sounds ‘bigger’. Repeat the test with a different instrument to learn why the same settings behave differently.
Compression and the rest of the mixing chain
Compression belongs within a larger gain structure. EQ placed before it changes which frequencies influence the detector, while EQ placed after it changes the final tonal balance. Saturation, transient shaping and automation can interact with compression, so processor order is part of the creative decision.
For related reading, start with The Decibel Explained, Gain Staging Explained, What Is EQ? and What Is Signal Flow?. The complete sequence is in the Music Production Reference Library.