Compression in practice: when and how to use a compressor
Moving beyond the controls — how to hear what a compressor is doing, when to compress and when not to, and settings for common sources.
Short answer
A compressor reduces the dynamic range of a signal — bringing loud peaks closer to quiet passages. In mixing, it is used to control dynamics (making a vocal sit consistently in the mix), shape transients (controlling the attack and sustain of drums), add density and glue (bus compression bringing a mix together), and create stylistic effects (heavy pumping compression in EDM). The most important skill is learning to hear compression — recognising when gain reduction is active, whether it is working with or against the source, and when a source would benefit from less compression rather than more. The controls (threshold, ratio, attack, release, makeup gain) are covered in the compression basics guide; this guide focuses on applying them.
Learning to hear compression
Most producers starting out compress everything because that is what professionals do — without being able to hear whether it is helping or hurting. Training your ears to hear compression is the prerequisite for using it well:
- Bypass A/B test: always compare the compressed and uncompressed signal at matched loudness (not with the makeup gain adding level — that artificially makes the compressed version sound better). The question is: does the compressor improve the source in its context, not just make it louder?
- Listen for transient change: when you engage compression with a slow attack, the initial attack of a sound passes through unaffected and the body is reduced — this adds punch and presence. With a fast attack, the transient itself is caught and reduced — this smooths and thickens the sound but can remove punch. Learn to distinguish these two effects.
- Listen for the release: the release time determines how the compressor lets go after gain reduction. Too fast a release causes pumping or "breathing" (audible level jumps). Too slow a release compresses the next note before the compressor has fully released from the previous one, creating a squashed, lifeless quality. The correct release is one you cannot hear — the compressor releases between notes or phrases.
- Watch the gain reduction meter: aim for 3–6 dB of gain reduction for transparent control; 6–12 dB for audible shaping; more than 12 dB is a stylistic effect. If you are regularly seeing 15–20 dB of gain reduction, the threshold is probably too low.
When to compress — and when not to
Compression helps when:
- A vocal performance has inconsistent level — some phrases much quieter, others much louder. Compression (or automation first, then light compression) brings these into a consistent range.
- A bass guitar or bass synth part is uneven — individual notes varying in volume in a way that disrupts the groove.
- You want to extend the sustain of a drum hit or plucked string — a slower attack, longer release allows the body of the sound to come forward after the transient.
- You are trying to glue a bus (drum bus, mix bus) together — moderate compression with a slow attack and medium release creates cohesion.
- You need a source to sit further back in the mix — heavier compression reduces dynamic variation, making a source feel more distant and "glued" to the track.
Do not compress when:
- The source is already dynamically consistent — adding compression adds noise and removes life without benefit.
- The source needs its transients intact for impact — a lightly recorded acoustic guitar, a pizzicato string, a tambourine. Compression can kill these.
- You are not sure what you are trying to achieve. Compression without intent makes mixes duller. Identify the problem first, then use the tool.
The controls and what they do
Brief reference (full detail in the compression basics guide):
- Threshold: the level above which the compressor acts. Lower threshold = more signal compressed = more gain reduction. Start high and bring it down until the meter shows 3–6 dB of gain reduction on peaks.
- Ratio: how aggressively the signal above the threshold is reduced. 2:1 is gentle; 4:1 is moderate; 8:1 is heavy. Start at 4:1 and adjust.
- Attack: how quickly the compressor responds. Slow attack (50–100 ms) lets transients through — adds punch. Fast attack (0.1–1 ms) catches transients — smooths and thickens. For drums, try slow attack first; for vocals, a medium attack (5–20 ms).
- Release: how quickly the compressor lets go. Set it so the gain reduction meter returns to zero between the main events — between drum hits, between vocal phrases. If the meter never fully recovers, the release is too slow.
- Makeup gain / output: compensate for the gain reduction so the compressed signal matches the uncompressed level for comparison. Do not use makeup gain to make the compressed version louder — this biases the A/B comparison.
Settings for common sources
These are starting points — adjust to taste and context:
- Kick drum: fast attack (1–5 ms), fast release (50–80 ms), ratio 4:1–6:1. Goal: control the tail without removing the initial punch. Or try slow attack to emphasise the transient.
- Snare: medium attack (5–20 ms), medium release (80–150 ms), ratio 4:1. Preserves the snap while controlling the room tail.
- Vocals: medium attack (5–20 ms), medium release (100–200 ms), ratio 3:1–4:1. Goal: even level with natural-sounding dynamics. Use automation first for large level variations.
- Bass guitar / synth bass: medium-fast attack (5–15 ms), medium release (100–200 ms), ratio 4:1–6:1. Goal: consistent note level and a controlled low-frequency footprint in the mix.
- Acoustic guitar: slow attack (30–80 ms), medium release (100–150 ms), ratio 2:1–3:1. Light compression preserving the pick attack and natural dynamics.
- Mix bus (master bus): slow attack (30–50 ms), slow release (200–400 ms), low ratio (1.5:1–2:1), threshold set for 1–3 dB of gain reduction. Goal: glue and cohesion, not obvious compression.
Serial and parallel compression
Serial compression is using two compressors in sequence. The first handles large, fast peaks (a limiter or fast compressor). The second shapes the overall dynamic contour more gently. The first protects the second from extreme signals; the second provides musical dynamic control. Common on vocals and drum buses.
Parallel compression (New York compression) blends an uncompressed dry signal with a heavily compressed wet signal. The dry signal preserves the transients and dynamics; the heavily compressed signal adds density and sustain underneath. The blend control determines how much compression character versus natural dynamics. This gives the best of both: punch from the dry signal, thickness from the compressed signal. Most DAWs support this via a blend knob on the compressor or through a parallel send.
For how compression interacts with sidechain sources, the sidechain compression guide covers the kick-bass relationship and ducking techniques.
Next step
The foundational controls are in the compression basics guide. For how compression fits into the full signal chain at mastering stage, the mastering basics guide covers limiting, loudness and final compression decisions. For bus compression specifically, EQ basics explains how to prepare each element before it reaches the bus compressor. For academic and practitioner depth, the Sound On Sound magazine archive has extensive compression technique articles, and AES publishes technical research on dynamic range processing.
Frequently asked questions
A compressor reduces gain above a threshold by a ratio set by the user — for example, a 4:1 ratio means for every 4 dB the signal rises above the threshold, only 1 dB passes through. A limiter is essentially a compressor with a very high ratio (10:1 or higher, often ∞:1) and a fast attack — it prevents the signal from exceeding the threshold at all. In mastering, a limiter is used as the final insert to prevent digital clipping. In mixing, a compressor with a moderate ratio is used to shape dynamics; a limiter is used to catch peaks.
Gain reduction is the amount by which the compressor is turning the signal down. If the meter shows −6 dB of gain reduction, the signal above the threshold is being attenuated by 6 dB. The makeup gain control (or output gain) then raises the overall level after compression, so the compressed signal can be as loud as the original — but with a narrower dynamic range (the loudest peaks are now closer to the quietest passages). This is how compression makes things sound louder and more present even though it is technically reducing gain.
Squashed or lifeless compression is usually caused by too much gain reduction (too low a threshold), too slow an attack (compressor reacts to the average level rather than peaks), or over-compression on the master bus. The transients — the attack phase of drums, picked notes, consonants on vocals — give music its sense of energy and punch. When compression removes those transients, the music loses life. The fix: raise the threshold (less gain reduction), use a faster attack to catch peaks specifically, and use a slower release to let the track breathe.
These refer to the circuit type of the compressor, which affects how it sounds. VCA (Voltage Controlled Amplifier) compressors (like the SSL G Bus) are fast and clean — good for drums, buses, mastering. FET (Field Effect Transistor) compressors (like the UA 1176) are fast with a punchy character — particularly suited to drums and rock vocals. Optical compressors use a light source and photocell — slower, musical, very natural-sounding on vocals and bass. Tube compressors are warm and colourful. Plugin emulations of all these types are widely available and produce similar tonal characters.
A sidechain is a secondary signal that controls the compressor's gain reduction rather than the signal being compressed. The most common use is sidechain compression on a bass track triggered by the kick drum: when the kick hits, the compressor detects it via the sidechain and reduces the bass level momentarily, creating "pumping" space for the kick. This technique, popularised by EDM and electronic music, allows the kick and bass to coexist in the low-frequency spectrum. The MusicTheGuide sidechain compression guide covers this in detail.