A MusicProductionWiki Publication Sound Better →
The Producer's Bible
All entries →
Intermediate
Understand first: Gain Staging Dynamic Range Headroom

Compression

noun / signal processing tool · kəm-PREH-shən
Every record you've ever loved was shaped by compression you couldn't hear — and that invisibility is the entire point.
Quick Answer

Compression reduces the dynamic range of audio by turning down signals that exceed a set threshold by a defined ratio, then using makeup gain to restore level — controlling peaks and shaping transients.

New to compression? Start here
Parameters Before / After Quick Reference Common Mistakes
Common Misconception

Most producers believe that adding compression always makes audio sound better, and that more compression means a more professional, polished result.

This is wrong. What makes compression appear to improve everything is almost always the makeup gain inflating the output level, not the gain reduction itself. In a scientifically controlled test — equal RMS levels, blind evaluation — many signals sound demonstrably worse with compression applied incorrectly than without it. The professional use of compression is about applying the minimum amount of gain reduction needed to solve a specific problem while preserving or enhancing the natural dynamics of the performance.

What Is Compression?

Compression is the process of automatically reducing the gain of an audio signal when it exceeds a defined threshold level, then restoring the average level with makeup gain. The result is a narrower dynamic range: the loudest moments are brought down, and the overall level is pushed up so that the quieter moments become more audible. In practice, this means a vocal that swings 20 dB between a whisper and a belt is tamed to a swing of 8 dB — present in the mix at every moment rather than lurching from inaudible to overwhelming. The compressor does not make audio louder in the sense of adding energy. It redistributes existing dynamic energy more evenly across time.

The perceptual effects of compression extend far beyond simple dynamic control. Because the attack and release time constants govern exactly when gain reduction engages and disengages relative to a transient, the compressor is simultaneously a transient-shaping tool. A slow attack on a snare drum lets the initial crack — the highest-amplitude, fastest-rising part of the signal — pass through unaffected before the compressor clamps down on the sustain and room. That choice transforms a flat, one-dimensional hit into a sound with snap and depth. Conversely, a fast attack on that same snare softens the crack, pushing the sustain forward in the mix. Neither setting is correct in isolation. Both are decisions about what the listener hears first and what they hear next.

Compression also functions as a glue tool at the mix and mastering stage. When all elements of a drum bus, or an entire mix, pass through a single compressor, the gain reduction responds to the combined signal. Loud moments in the kick cause slight gain reduction across the snare, hi-hats, and room — binding the kit into a coherent physical event rather than a collection of independent sounds. This phenomenon, often called bus compression, is one of the most underappreciated techniques in professional mixing.

There are five compressor topologies in common use — VCA, optical, FET, variable-mu (vari-mu), and digital — each with distinct character arising from how its gain reduction circuit responds to the incoming signal. A VCA compressor like the SSL G-Bus or API 2500 responds with precision and speed, making it the standard choice for drum busses and mix busses where control is paramount. An optical compressor like the LA-2A uses a light source and photocell whose physical properties introduce a natural, program-dependent release that feels organic on vocals and bass. A FET compressor like the UA 1176 has a near-instantaneous attack and an aggressive, forward character that has defined rock vocals and drum room mics for sixty years. Understanding which topology to reach for before touching a knob is the difference between a mix that sounds engineered and one that sounds produced.

"The best compression is when you don't hear it. When you hear compression, something has gone wrong."

— Rick Rubin, Producer (Johnny Cash, Red Hot Chili Peppers, Adele) · The Creative Act: A Way of Being

Compression reduces dynamic range by attenuating signal above a set threshold, then restoring level with makeup gain — shaping transients, controlling peaks, and gluing elements into cohesion when used at the right ratio with the right time constants.

How Compression Works

At its core, a compressor contains two functional paths: a gain reduction element and a detector (also called a sidechain or control path). The audio signal passes through the gain reduction element — a VCA, optical cell, FET, or tube stage — which is capable of attenuating the signal by a variable amount. The detector continuously measures the signal's amplitude, compares it to the threshold setting, and sends a control voltage to the gain reduction element telling it how much to attenuate. When the signal is 10 dB above the threshold and the ratio is set to 4:1, the detector tells the gain reduction element to attenuate by 7.5 dB, allowing only 2.5 dB of the excess level to reach the output.

The attack and release time constants control how fast the control voltage changes in response to threshold crossings. Attack time is defined, in most compressors, as the time required to reach 63% of the full gain reduction — not 100%. This means that a 10 ms attack setting does not fully engage the compressor until 20–30 ms after threshold crossing for typical program material. This matters enormously in practice: a 10 ms attack on a kick drum at 120 BPM allows roughly 10–15 ms of uncompressed transient to pass before gain reduction reaches its working level. That window is what producers call the transient punch — the initial click and crack of the instrument that carries the perception of impact.

Makeup gain — the output gain stage applied after gain reduction — is the mechanism by which compression produces its most seductive perceptual effect: loudness. A compressor applying 6 dB of gain reduction on peaks, followed by 6 dB of makeup gain, has not changed the peak level at the output. But the average level is now higher, because the valleys are not pulled down while the makeup gain raises everything. The correct evaluation method is to null the output level before bypassing — set makeup gain to 0 and listen at matched RMS, not matched peaks.

A detector circuit measures the signal against the threshold and sends a control voltage to a gain reduction element; attack and release time constants govern when that reduction engages and disengages, making the compressor simultaneously a dynamic and a transient-shaping tool.

Compression — Key Parameters

Threshold

The dB level at which the compressor begins to act. Signal above the threshold is subject to gain reduction; signal below passes through unchanged. Lower threshold = more of the signal gets compressed. Setting threshold at -10 dBFS on a vocal catches only the peaks; setting it at -30 dBFS means the compressor is working almost constantly.

Ratio

The ratio of input level change to output level change above the threshold. At 4:1, a signal 8 dB above threshold emerges only 2 dB above it. At infinity:1 (limiting), no signal exceeds the threshold. Low ratios (1.5:1–3:1) feel transparent and musical. High ratios (8:1–20:1) clamp the signal visibly, creating the pumped, controlled sound.

Attack

The time constant (ms) for the compressor to reach full gain reduction after the signal crosses the threshold. Fast attack (0.1–5 ms) clamps transients immediately — drums lose crack but peaks are controlled tightly. Slow attack (20–50 ms on a snare) lets the initial transient through before gain reduction kicks in, preserving snap and punch. This is the primary transient-shaping parameter.

Release

The time constant (ms or s) the compressor takes to return to unity gain after the signal drops below the threshold. Too-fast release (under 30 ms on a mix bus) causes distortion artifacts as gain pumps at audio frequencies. Too-slow release (over 800 ms on a drum bus) means the compressor never fully recovers between hits, reducing perceived punch of the next transient.

Knee

Controls how abruptly the compression ratio is applied as the signal approaches and exceeds the threshold. Hard knee: full ratio applied immediately at the threshold — sounds more aggressive and obvious. Soft knee: ratio is gradually applied over a dB range (typically ±3 to ±6 dB) centered on the threshold — smoother and more transparent.

Makeup Gain

A fixed gain stage applied after the compressor's gain reduction circuit to compensate for the level lost during compression. Setting it incorrectly inflates output levels and makes the result appear louder during A/B comparison. Always null your output level when evaluating compression decisions. This principle is fundamental to gain staging.

Lookahead

A digital-only feature that delays the audio path while allowing the detector circuit to analyze the signal in advance. This allows the gain reduction to begin before the transient arrives at the output, effectively achieving zero attack time without the artifacts of truly instantaneous gain reduction. Used in mastering limiters.

Threshold sets where compression starts, ratio sets how much gain reduction is applied, attack and release determine the temporal shape of the compression, knee controls transition smoothness, and makeup gain restores level — each parameter changes a different sonic dimension of the result.

Quick Reference Card

4 dBMaximum gain reduction on any mix bus before the compressor is working against you

Applies to mix bus and mastering bus compression in any genre; break this rule only when deliberate pumping is an aesthetic choice, as in sidechain-driven house or techno.

ParameterDrumsVocalsBassBus
Threshold-18 to -25 dBFS-15 to -20 dBFS-15 to -22 dBFS-15 to -20 dBFS
Ratio4:1 – 8:12:1 – 4:13:1 – 6:12:1 – 4:1
Attack5–15 ms (snare), 0.1–3 ms (kick control)10–30 ms30–80 ms20–50 ms
Release40–80 ms60–120 ms80–200 ms100–300 ms (auto often best)
Gain Reduction6–10 dB3–8 dB4–8 dB2–4 dB

Use these as starting points, set threshold first until GR reads the values in row 5, then adjust attack and release to taste by listening for transient punch and breathing artifacts.

Share

⚡ Gain Reduction Calculator

Enter your settings to see exact gain reduction applied above the threshold.

— dB gain reduction applied
above threshold
output level
after makeup gain
Free via The Producer's Briefing

Signal Chain Position

Signal chain position of Compression in music production Preamp mic / line / instrument gain structure EQ (corrective) high-pass, notch filters problem frequencies Clip Gain gain staging, trim level before compressor COMPRESSION threshold · ratio · attack release · makeup gain ◀ YOU ARE HERE EQ (additive) tone shaping, air after dynamics stable Saturation harmonic color, glue tape / tube / console Reverb / Delay time-based effects space and depth Limiter peak control final ceiling
Preamp
mic / line / instrument gain structure
EQ (corrective)
high-pass, notch, problem frequencies
Clip Gain / Trim
gain staging before compressor
Compression
threshold · ratio · attack · release · makeup gain
▶ You are here
EQ (additive)
tone shaping, air, after dynamics stable
Saturation
harmonic color, tape / tube / console
Reverb / Delay
time-based effects, space and depth
Limiter
peak control, final ceiling

Compression before EQ controls the dynamic range of the raw signal; EQ after compression shapes the tonal character of the now-stabilized signal — running EQ first then compressing risks the compressor reacting differently to boosted frequencies than to the original source.

Interaction Warnings

  • Compression interacts problematically with reverb when the compressor is placed after the reverb return in the signal chain: the reverb tail triggers continued gain reduction after the dry signal ends, causing the reverb to pump and swell audibly in the gaps between notes. Fix: always compress the dry signal before the reverb send.
  • Heavy compression (above 6 dB GR) on a bass instrument interacts problematically with low-pass filter resonance or additive low-frequency EQ boosts placed after the compressor. Fix: apply corrective EQ and resonance control before the compressor.
  • Sidechain compression triggered by a kick drum interacts problematically with a bass synthesizer whose low-end fundamental is below 40 Hz when the sidechain is not high-pass filtered. Fix: high-pass the sidechain input at 60–80 Hz so the compressor responds only to the audible body of the kick.

Diagram

Compression: Input vs. Output Level with Threshold and RatioCompression: Input vs. Output LevelInput Level (dB)Output Level (dB)1:1 (no compression)4:1 ratio above thresholdThresholdKnee point

This diagram shows how compression works: when the input signal rises above the threshold, the output increases at a reduced rate set by the ratio. Below the threshold, input equals output (no gain reduction). The knee point marks where the ratio engages. Understanding this input/output relationship is the foundation of every compression decision.

History of Compression

Origins: Broadcast Utility (1930s)

Western Electric engineer Harold Black developed feedback amplifier theory in the late 1920s at Bell Labs, laying the conceptual groundwork for automatic gain control — the technical ancestor of audio compression. The first practical broadcast compressors emerged in the 1930s as purely utilitarian devices: NBC and CBS engineers in New York needed to prevent transmitter overload when announcers leaned into microphones or orchestras swelled unexpectedly. No one was thinking about what these devices did to the feel of a performance; the goal was to stop the catastrophic failure of expensive RF equipment.

Compression as Creative Tool (1950s–60s)

The creative application of compression as a deliberate sonic choice began in earnest with engineer Bill Putnam Sr. at Universal Recording in Chicago in the late 1940s and early 1950s. The UA 175B tube compressor, which Putnam's company produced, was a variable-mu design that added warmth and density to a signal in a way that was audibly musical rather than merely functional. By the time Putnam was recording Nat King Cole and Frank Sinatra at United Western in the mid-1950s, compression had become a creative decision — which unit to use, how hard to push it, and how its particular topology colored the sound were all choices made in the service of the performance.

The 1176 and the SSL Era (1967–1981)

The 1176 Peak Limiter, brought to market in 1967, defined a generation of rock and pop recording. Its 4-button 'all-buttons-in' mode became a standard drum room technique. John Bonham's room sound on Led Zeppelin's 1971 recordings at Headley Grange was shaped partly by 1176s on the room microphones — the way those compressors responded to the enormous transients in that stairwell is central to why that drum sound still sounds physically impossible fifty years later.

The SSL G-Bus and Modern Mixing (1981–Present)

The SSL G-Bus compressor, introduced on the SSL 4000 G Series console in 1981, fundamentally changed what bus compression meant. Engineers at facilities like Abbey Road discovered that running the bus comp continuously at 2:1 or 4:1, with threshold set to achieve roughly 2–4 dB of gain reduction on peaks, transformed the way a mix held together. The SSL G-Bus compressor clone is now the most widely emulated piece of hardware in software history, present in essentially every major DAW's built-in processing suite.

Compression moved from a defensive broadcast tool in the 1930s to a creative production instrument through engineers like Bill Putnam Sr. in the 1950s, and was codified as a standard studio technique by the UA 1176 in 1967 and the SSL G-Bus compressor in 1981.

How Producers Use Compression

The built-in Glue Compressor models SSL G-Bus behavior directly. Use it on the drum bus with Peak detector, 4:1, attack at 0.5 ms, release auto, ceiling 0 dB for classic bus glue. The regular Compressor's Auto release button is reliable enough for vocal chains at 3:1 with a 20 ms attack.
The Vintage VCA in Logic's compressor suite directly models the SSL G-Bus. The Vintage Opto models the LA-2A behavior — correct starting point for vocal compression. Set Vintage Opto to Compress mode, Comp knob at roughly 50%, Peak Reduction at 40–60% — the program-dependent behavior handles release automatically.
The Fruity Compressor is most usable with Type switch in Peak mode for drums. For bus compression, the Mixer's built-in Peak Controller routing is the correct approach. The Parametric EQ 2's sidechain input enables frequency-dependent compression when routed correctly.
The BF-76 (bundled) is a reliable 1176 model for drums and vocals. For mix bus, the SSL G-Bus Compressor plugin (Waves) is the standard. Pro Tools' stock Dynamics III compressor is serviceable as a utility tool but has no meaningful character — use it for gain control only, not for color.

Drums: On individual drum tracks, compression is primarily a transient-shaping and sustain-control tool. On kick drum, the standard starting point is a fast-to-medium attack (5–15 ms) to preserve the initial click and beater impact, a ratio of 4:1 to 6:1, and a release set just fast enough to recover before the next kick hit. On the drum bus, a VCA compressor (SSL G-Bus style) set at 2:1 to 4:1 with a medium attack (10–30 ms) and an auto or 100–200 ms release, with threshold adjusted for 3–6 dB of gain reduction on the loudest hits, will glue the kit into a single physical event.

Vocals and Bass: On lead vocals, the LA-2A optical approach — or any compressor with a program-dependent release — is the industry default because the release adapts to the natural cadence of speech and singing. Two-stage vocal compression — a first compressor at a low ratio (2:1) for general dynamic control, followed by a second compressor at a higher ratio (6:1 or more) for peak limiting — is standard practice on professional vocal chains.

Mix Bus: Mix bus compression should be set first — before all other mix decisions are made — so that you're mixing into the compressor. The classic SSL G-Bus setting: 4:1, attack 30 ms, release auto, threshold for 2–4 dB of gain reduction on the loudest sections. For mastering, a vari-mu compressor (Manley Vari-Mu style) at very gentle settings — 1.5:1 to 2:1, 1–3 dB of gain reduction — adds density and low-end weight without audible pumping.

"I use compression to make things feel right, not to make them louder. There's a difference and most people confuse the two."

— Bob Clearmountain, Mix Engineer (Bruce Springsteen, The Rolling Stones, Bryan Adams) · Tape Op Magazine Issue 78

On drums, compression shapes transients and glues the kit; on vocals, it creates consistency and presence; on the mix bus, it binds the full arrangement into a coherent, moving whole — and the correct tool for each context depends on matching compressor topology to the required time constants.

Compression by Genre

ParameterTrapHip-HopHouseRockMastering
Ratio8:1–20:1 on melodic elements for aggression6:1–10:1 on drums; 2:1–3:1 on 808s to preserve sub-fundamental4:1 on kick for punch; 2:1 on mix bus with fast release for intentional pump4:1–6:1 on room mics; 3:1 on individual drums; 2:1 on bus1.5:1–2:1 only; heavier ratios add distortion artifacts at mastering levels
AttackUnder 1 ms across entire mix — destroys all transient informationFast (1–5 ms) on 808 kick body; medium (15–30 ms) on snare to preserve crackSlow (50–100 ms) on main room to allow kick transient through, then compress sustainMedium (10–20 ms) on snare; slow (30–60 ms) on room and overhead micsSlow (30–80 ms) to preserve the attack of every instrument in the full mix signal
ReleaseUnder 30 ms on complex material — causes intermodulation distortion and pumping artifacts60–100 ms on individual drums; auto-release on bus to track the syncopated rhythmFast (30–60 ms) on mix bus — the pump on the off-beat is the effect, not a mistake80–150 ms on drum bus; 200–400 ms on vocal to allow phrase breathingAuto or program-dependent; manual release only when correcting a specific pumping problem
Gain Reduction (avg)Consistent 12+ dB on any source except deliberate limiting6–10 dB on 808 and kick; 3–5 dB on snare; 2–3 dB on bus3–6 dB on kick; 6–8 dB on bus if pumping is intentional6–12 dB on room mics; 4–8 dB on snare; 2–4 dB on drum bus1–3 dB; never more than 4 dB on a well-mixed source
Circuit TypeSame compressor on every source regardless of characterVCA for kicks and snares; optical on 808 for smooth sub controlVCA on kick and bus for punch; FET on percussion for biteFET (1176) on room mics and electric guitars; optical on bass and acoustic sourcesVariable-mu or optical for tonal glue; VCA for precise peak control before limiter
Share

Hardware vs Plugin vs Stock

AspectHardware OriginalTop PluginDAW Stock
Gain Reduction SpeedAnalog VCA and FET circuits have physical rise times; even at minimum attack settings the gain reduction ramps, not steps — which is part of why they sound musical at extreme settingsWaves CLA-76 and UAD 1176 emulations model the detector circuit's ballistics accurately at normal settings; diverge at All-Buttons-In mode where behavior is highly unit-specificDAW stock compressors are mathematically precise — the attack time is what you set, no more, no less — which reads as clinical to some ears
Harmonic CharacterTransformer-coupled hardware (Neve, SSL) adds low-order even harmonics naturally through the input and output transformers, independent of compression amountUAD Neve 33609 and SSL G-Bus emulations model the transformer saturation with reasonable accuracy; audible at high gain settings but subtler than hardware at nominal levelsNo harmonic coloration by design; purely linear gain reduction — pair with a saturation plugin post-compressor to approximate the character of transformer-coupled hardware
Stereo BehaviorHardware stereo compressors use a single detector feeding both channels from a summed side-chain; true stereo linkage maintains image integrity under gain reductionMost emulations reproduce linked stereo behavior; mid-side compression options in plugins like FabFilter Pro-C 2 allow frequency-selective stereo control unavailable on most hardwareLogic's Vintage and Ableton's Glue implement linked stereo correctly; unlinked mode in stock compressors can shift the stereo image under heavy gain reduction
LatencyZero latency by definition; no delay compensation required in the mixLinear-phase and lookahead compressors introduce latency of 1–20 ms; modern DAWs compensate automaticallyDAW stock compressors are typically zero or near-zero latency
Recall and AutomationComplete parameter recall requires manual note-taking or a recall sheet; no automation unless integrated with a hybrid control systemFull DAW automation of every parameter including threshold, ratio, attack, and release; snapshots and A/B comparison built into most modern plugin interfacesFull recall and automation native to the DAW; no licensing or compatibility concerns

Plugin Recommendations

Free
Analog Obsession LALA
Analog Obsession
TDR Kotelnikov
Tokyo Dawn Records
Airwindows CL
Airwindows
Mid ($)
CLA-76
Waves
Marquee VCA
Plugin Alliance
Glue Compressor
Cytomic (also in Ableton)
Pro ($$)
UAD 1176 Series
Universal Audio
Neve 33609
UAD / Softube

Before and After

Without Compression

An uncompressed lead vocal track has a measured dynamic range of approximately 18–22 dB between the quietest phrases (sitting around −30 to −28 dBFS) and the loudest moments (peaking at −10 to −8 dBFS). The loudest peaks stick out of the mix by 6–8 dB, requiring the overall vocal level to be set low enough that the quiet phrases are buried under guitars and synths. The performance is emotionally legible in solo but disappears and reappears in the full mix context.

With Compression

After applying an optical-style compressor (LA-2A emulation) at 3:1 ratio, threshold set at −20 dBFS for 5–8 dB of gain reduction on peaks, attack at 15 ms, program-dependent release, and makeup gain adjusted to null RMS level: the dynamic range narrows to 8–10 dB. The vocal now sits consistently in the mix through every moment of the song. Quiet, intimate phrases are now audible in the mix context. The listener's attention is directed to the words and the emotion rather than to the mechanics of the level.

The Producer's Briefing
The Producer's Briefing — practical technique, gear intel, no fluff.

Compression In The Wild

The BeatlesWhen I'm Sixty-Four (1967, Sgt. Pepper's Lonely Hearts Club Band). Produced by George Martin.
The bass guitar was recorded direct and compressed through an optical unit at EMI Studios. Every note sustains at the same level, giving it a synthesizer-like consistency — optical compression doing what it does best.
Led ZeppelinWhen the Levee Breaks (1971, Led Zeppelin IV). Produced by Jimmy Page.
Room microphones placed on the first-floor landing of Headley Grange stairwell, compressed with 1176s. The compressor's attack lets the initial transient escape before clamping down — then the release pulls up the room tail. The result is a geological event, not a drum recording.
Michael JacksonBillie Jean (1982, Thriller). Produced by Quincy Jones.
Engineer Bruce Swedien used tight VCA compression on the kick — every hit lands at identical volume with identical decay. The 5–10 ms attack lets the beater click through before the compressor engages. That window is the entire sonic identity of this kick.
Daft PunkOne More Time (2000, Discovery). Produced by Daft Punk.
The four-on-the-floor kick interacts with fast mix bus release times — the mix breathes down on every kick hit, then surges back in the gap. The release time is tuned to 120 BPM tempo. This is not a mistake — it was calibrated as a rhythmic and textural element.
Kendrick LamarHUMBLE. (2017, DAMN.). Produced by Mike Will Made-It.
Enormous low-end body, precisely controlled sub-bass tail, and a fast punchy transient at the front. The slow enough attack (approximately 5–10 ms) lets the beater click through before clamping the body aggressively — preventing sub-bass bleed while preserving the impact.
RadioheadKarma Police (1997, OK Computer). Produced by Nigel Godrich.
Beginning at approximately 2:35, the outro build demonstrates bus compression increasing coherence. As more elements layer in, the compressor works harder, bringing overall level down slightly while pushing average level upward — the mix gets denser, not just louder.
AdeleRolling in the Deep (2010, 21). Produced by Paul Epworth.
When the vocal enters at 0:08, the consonants of 'There's a fire' hit sharply before the compressor engages — slow attack optical compression preserving the intelligibility and natural character that make the performance feel immediate and unprocessed.

From the synthesizer-like bass on When I'm Sixty-Four to the intentional pumping of One More Time, compression in professional recordings is either completely invisible or deliberately weaponized as a rhythmic texture — rarely accidental, never incidental.

Compression vs Limiting

Compression uses a ratio — it reduces gain proportionally above the threshold and signal can still exceed it. Limiting uses a ratio of ∞:1 — no signal passes above the ceiling. Use compression for musical dynamic control; use limiting for transparent peak management and loudness maximization at the final stage.

Compression vs Saturation

Compression is gain reduction — it makes loud signals quieter in a controlled way. Saturation is harmonic distortion — it adds frequency content. Both can "glue" a mix, but through opposite mechanisms. Many professional chains use both: compression for control, saturation for color and density.

Types of Compression

VCA (Voltage Controlled Amplifier)

SSL G-Bus Compressor, dbx 160, API 2500

VCA compressors use a voltage-controlled amplifier in the gain reduction path, making them fast, precise, and punchy. Best for drums, drum busses, and any source where you want aggressive gain reduction without excessive coloration — this is the compressor that defined the sound of 1980s and 1990s rock and pop.

FET (Field Effect Transistor)

Urei 1176, Purple Audio MC77

FET compressors use a field effect transistor as the gain reduction element, allowing extremely fast attack times (down to 20 microseconds on the 1176) with a distinctive harmonic character. They impart a forward, aggressive quality that works especially well on vocals, room mics, and anything that needs presence and energy.

Optical

Teletronix LA-2A, Tube-Tech CL 1B

Optical compressors use a light-sensitive resistor (photocell) and a light source whose intensity tracks the input signal. The inherent lag of the optical element creates a program-dependent attack and release that responds musically to the dynamics of the source — ideal for vocals, bass, and acoustic instruments.

Variable-Mu (Tube)

Fairchild 670, Manley Variable Mu

Variable-mu compressors use tubes operating in their nonlinear region to achieve gain reduction. Attack and release times are inherently slow and program-dependent, making these compressors exceptional for gluing program material and adding harmonic richness. The Fairchild 670 stereo unit remains one of the most copied designs in mastering.

Multiband

Neve 33609 (sidechain EQ), Waves C6

Multiband compressors split the audio signal into frequency bands and compress each band independently. Experienced engineers use the minimum number of bands required to solve a specific problem, not as a default mixing tool.

Parallel (New York)

Any compressor in a parallel configuration

Not a compressor circuit type but a routing technique: the dry, uncompressed signal is blended with a heavily compressed version. The result retains the natural transients and dynamics of the dry signal while the compressed signal adds density, sustain, and apparent loudness underneath.

VCA compressors provide speed and precision for drums and busses; optical compressors provide musical, program-dependent response for vocals and bass; FET compressors provide aggressive forward character for rock and hip-hop; variable-mu compressors provide warmth and density for mastering.

The Producer’s Verdict

Compression is the most misunderstood tool in production — not because it’s complex, but because its most seductive effect (perceived loudness from makeup gain) disguises bad decisions as good ones. Master the null test before anything else: bypass your compressor and match output levels by RMS, not peaks. If the compressed version doesn’t improve on that honest comparison, your settings are wrong.

The Starting Point 4:1 ratio · 10ms attack · 150ms release · null makeup gain Set threshold for 4–6 dB of gain reduction on peaks. Use this on every source until you can hear exactly what each parameter changes before deviating from it.
Topology Decision VCA for drums · Optical for vocals · FET for aggression · Vari-mu for mastering Circuit type determines character, not just speed. A VCA on a vocal and an optical on a kick are both technically correct settings that will sound wrong. Match topology to source first, then dial parameters.
Mix Bus Rule Set it first. Mix into it. Never add it last. SSL G-Bus style: 4:1, attack 30ms, auto release, 2–4 dB GR on peaks. A compressor added after the mix is finished is a different tool than one the mix was built around. The latter glues. The former patches.
Transient Control Attack time is the most important parameter Ratio affects how much gain reduction. Attack affects what the listener perceives as impact. A slow attack on a snare (20–30ms) lets the crack through before the compressor clamps down. That window — not the ratio — is what makes a snare sound like a snare.
When NOT to Compress When the performance dynamic IS the expression A whispered vocal rising to a shout. A piano phrase that breathes. A bass line that swells intentionally. Not every dynamic excursion is a problem to solve. Before compressing anything, ask whether the dynamic range you’re about to eliminate is the performance. Sometimes the answer is yes.
The Professional Standard Maximum 4 dB GR on any mix or mastering bus Beyond 4 dB on a mix bus, you are no longer gluing — you are fighting. The mix is not ready for that much compression. Fix the gain staging, fix the arrangement density, or fix the individual track levels before reaching for more bus compression.

The producers who use compression best are the ones who use the least of it to achieve the most. Every dB of unnecessary gain reduction is a dB of dynamic life you’ve chosen to remove. The invisibility of great compression is not an accident — it is the entire discipline.

Common Mistakes with Compression

Killing the Transient With Too-Fast Attack

Symptom: Drums sound flat, distant, and lifeless. The kick loses its click and beater impact; the snare has no crack.

Cause: Attack time set below 2 ms on drum tracks or a drum bus. At 0.1–1 ms, the compressor engages before the transient has time to reach the output.

Fix: Set attack to 10–30 ms on snare, 5–15 ms on kick. Increase attack time by 5 ms increments until the crack and punch return.

Pumping and Breathing on the Mix Bus

Symptom: The entire mix surges and dips in level in time with the kick drum. Every kick hit is followed by an audible swell as the rest of the mix comes back up.

Cause: Release time set too fast on the mix bus compressor — typically under 80 ms.

Fix: Increase release time to 150–400 ms on the mix bus. Use auto-release if available; it adapts to program material and eliminates the need to calculate tempo-matched release times.

Evaluating Compression With Mismatched Levels

Symptom: Every compressor sounds like it's improving the signal — but the mix sounds over-compressed and lacking in dynamic life.

Cause: Makeup gain is inflating the output level above the dry signal level. Louder almost always sounds better in a short A/B comparison — this is a psychoacoustic bias, not a sonic judgment.

Fix: Before evaluating any compression decision, null the output level. Reduce makeup gain until the compressed signal measures at the same RMS level as the unprocessed signal. Perform the A/B at matched levels.

Over-Compressing Vocals Into Lifelessness

Symptom: The vocal sits at a perfectly consistent level but feels mechanical, robotic, and emotionally flat.

Cause: Threshold set too low and ratio set too high, so the compressor is working 15–20 dB of gain reduction throughout the entire phrase. The dynamic range of the performance — the very thing that communicates emotion — is compressed out of existence.

Fix: Set threshold so the compressor is working no more than 6–10 dB on the loudest peaks. Lower the ratio to 3:1 to 4:1. Use clip gain or volume automation before compression to tame the worst peaks by hand.

Slow Release Killing Punch on a Drum Bus

Symptom: The drum bus has consistent average level but every hit sounds soft and indistinct. The kit sounds like it's being played through a pillow.

Cause: Release time set above 800 ms on the drum bus compressor. The compressor applies gain reduction in response to each hit but does not recover fully before the next hit arrives.

Fix: Calculate the tempo-relative release. At 120 BPM, kick hits arrive every 500 ms; set release to 200–350 ms so the compressor recovers at least 80% of gain reduction between hits.

Using Makeup Gain as a Loudness Crutch

Symptom: The mix gets louder after compression but the headroom is gone. The master limiter is working 6–10 dB of limiting, introducing artifacts and smearing transients.

Cause: Makeup gain on multiple compressors across the session is adding cumulative level gain. Each compressor adds 3–6 dB of makeup gain; by the time the signal reaches the master limiter, the true peak is already high.

Fix: Audit every compressor in the session and confirm that makeup gain is compensating for the actual gain reduction applied, not inflating beyond it. Keep the mix bus output at −6 to −3 dBFS headroom before the master limiter.

The most common compression errors are killing transients with too-fast attack, creating audible pumping with too-fast release, and mistaking makeup gain inflation for sonic improvement — all three sound different, have specific identifiable causes, and have specific numerical fixes.

Red Flags and Green Flags

🔴 Red Flags

  • The mix sounds flat and two-dimensional with no sense of depth — threshold is too low and ratio too high. Raise threshold until GR averages 2–4 dB only on peaks.
  • The snare drum sounds soft and pillowy with no attack or crack: attack time is under 3 ms. Increase attack to 15–25 ms.
  • The vocal sounds like it is breathing or gasping between phrases — release time is too fast (under 50 ms). Slow release to 100–200 ms or engage auto-release.
  • The low end of the mix is distorting subtly — release time on the mix bus compressor is under 30 ms, causing the gain reduction to pump at sub-audio frequencies. Increase mix bus release to 100 ms minimum.
  • The mix A/Bs loudly against the uncompressed version but loses energy when level-matched: excess makeup gain is inflating the perceived quality. Null the output before A/B comparison.

🟢 Green Flags

  • The gain reduction meter on the drum bus moves rhythmically with the groove — 4–6 dB on kick hits, recovering cleanly between beats — the compressor is breathing with the track rather than fighting it.
  • A vocal phrase that previously ranged from -18 dBFS on quiet notes to -3 dBFS on belted notes now sits consistently between -12 and -6 dBFS throughout.
  • Bypassing the compressor on the mix bus causes the mix to feel looser and slightly chaotic, and elements feel less like they belong together — the bus compression is providing cohesion (2–3 dB GR) without becoming the loudness tool.
  • The parallel compression blend on the drum bus adds body and density when the fader is raised but the attack and crack of the snare do not diminish — the dry and compressed paths are correctly balanced.
  • After compression and level-matched A/B comparison, the compressed version of a bass guitar sounds more consistent in level across the fretboard — the compressor is controlling the natural acousto-mechanical inconsistency of the instrument.

Your Progression with Compression

Beginner

Before touching a compressor, identify the dynamic issue. Start with the simplest possible setting: 4:1 ratio, medium attack (20 ms), medium release (200 ms), threshold adjusted until you see 4–6 dB of gain reduction on the loudest moments, and makeup gain set to zero. Then raise the makeup gain until the compressed version is at the same perceived loudness as the bypass. You are now hearing what the compression is actually doing to the sound, not what the louder level is doing to your perception of the sound.

Intermediate

Once you can hear gain reduction reliably at matched levels, take a snare drum with no compression. Set a compressor to 6:1, threshold so it's working 8 dB of reduction, and sweep the attack from 0.5 ms to 50 ms while the track plays. At 0.5 ms, the crack disappears. At 50 ms, the crack comes through fully and the compressor catches only the sustain. Everything between those extremes is a decision about how much transient energy you want in the mix. This is where compression goes from a dynamic control tool to a sound-design tool.

Advanced

When you are ready for professional results, work in series: a slow optical compressor (LA-2A style, 2:1, program-dependent release) before a faster VCA compressor (4:1, 10 ms attack, 150 ms release) on a vocal chain. The optical unit handles the macro dynamic variation across the performance while the VCA handles the micro variation within phrases. Set your mix bus compressor before beginning a mix and mix every element into that compressed bus — your balancing decisions will be calibrated to the final, glued sound from the start.

Frequently Asked Questions

A compressor monitors the level of an audio signal and automatically turns it down whenever it rises above a set threshold, by a ratio you specify. When a vocal hits a loud note 8 dB above threshold with a 4:1 ratio, the output rises only 2 dB above threshold instead of 8 dB — the dynamic range is reduced. The makeup gain stage then raises the overall level, which means the average level is louder and the performance sounds more consistent.

The clearest audible symptom is the sound losing its sense of depth and three-dimensionality — everything feels flat and equally loud, with no sense of perspective between near and distant elements. Metering tells part of the story: consistent gain reduction of more than 8–10 dB on a single source or more than 4 dB on a mix bus is usually too much. The diagnostic test is to bypass the compressor — if the uncompressed version sounds more alive and dynamic while still being usable, you are working too hard.

Use series compression when you need to control dynamic range for a technical reason — a vocal that clips downstream processors, a bass that moves too much in the mix, a drum bus that has erratic peaks. Use parallel compression when the source already has great dynamics you want to preserve but needs more density and weight underneath.

The 1176 (FET) has attack times as fast as 20 microseconds and responds aggressively to transients — it adds presence and forward energy. Use it on anything that needs to be pushed to the front of a mix: lead vocals, room mics, electric guitars. The LA-2A (optical) has a program-dependent, inherently slow response. Use it on sources that need transparent density without an audible compressor sound: bass, acoustic guitar, smooth vocal styles. Both on the same vocal in series — 1176 first for peak control, LA-2A second for density — is a combination used on an enormous number of major-label records.

Attack time determines how much of the initial transient passes through before gain reduction engages. A fast attack (1–3 ms) clamps the transient immediately, reducing the crack of a snare hit. A slow attack (15–30 ms) lets the full transient through before the compressor reacts, preserving the snap and perceived punch of the hit. For a snare drum you want presence in the mix, start at 15–20 ms attack, 60 ms release, and 4:1 ratio.

Sidechain compression routes an external signal to the compressor's detector circuit, so the gain reduction is triggered by one signal but applied to another. The most practical application is ducking: the kick drum signal is fed to the sidechain of a compressor on the bass guitar, so every time the kick hits, the bass ducks by 2–3 dB, creating space in the low end without EQ carving. This is necessary — not optional — in dense electronic and hip-hop productions where the kick and bass occupy the same sub frequencies.

Because makeup gain is masking the damage. Null your output: set the compressor's output lower to ensure you are evaluating compression decisions at equal perceived loudness. Mix bus compression should be set to no more than 2–4 dB of gain reduction, with a ratio of 2:1 or lower, and attack times of 30–50 ms to preserve the transient content of your drums.

Compression at mastering is used for tonal shaping, cohesion, and gentle dynamic control — typically 1–2 dB of gain reduction at a 2:1 ratio with slow attack and auto release, applied before the limiter. Limiting is used for peak control and final loudness maximization. They cannot replace each other: compression shapes the body of the signal and determines how the limiter will subsequently react to transients, while the limiter handles the final ceiling.

Tools for This Entry

⚡ Gain Reduction Calculator

Interactive Tool

Calculate exact gain reduction for any compressor setting. Shows output level and final level after makeup gain.

Jump to Calculator ↓
More tools coming: Attack/Release visualizer, parallel blend calculator, sidechain frequency finder. Subscribe to get notified.

Also in The Bible

Last Verified: May 2026 · 2026 Edition · Part of The Producer's Bible

Keep Going

Sound better by Friday.

One email a week. The techniques behind the terms — what's actually happening in the records you love, and how to apply it in your session today.

← Browse The Bible Techniques → Gear Comparisons →