A multiband compressor splits a signal into several frequency bands with crossover filters, compresses each band with its own settings, then adds the bands back together. The low end can be held steady while the highs are left alone, or harsh upper mids can be tamed only when they flare up, without the rest of the sound being pulled down with them.

That’s the key difference from a normal, single-band compressor: when a loud bass note hits a single-band compressor, everything gets turned down, cymbals and vocals included. A multiband compressor confines the reaction to the band that caused it. It’s powerful, widely used in mastering and broadcast, and easy to overdo.

What is a multiband compressor?

A multiband compressor (also called multiband dynamics or split-band compression) has three stages:

  1. Crossover: filters divide the signal into adjacent bands, typically three to five (low, low-mid, high-mid, high and so on).
  2. Per-band dynamics: each band has its own threshold, ratio, attack, release and makeup gain, and its own detector, so it reacts only to its own content.
  3. Summing: the processed bands are added back together into one output.

Because each band’s gain changes with its level, a multiband compressor works like an equalizer that moves on its own: it turns a frequency range down only when that range gets loud. Many modern units also offer upward compression and expansion per band, which makes them general-purpose multiband dynamics processors. The extreme OTT setting is a well-known example of combined upward and downward multiband compression.

History

Splitting audio into bands for dynamics processing began with noise reduction. Dolby’s A-type system, introduced in 1965, divided the signal into four bands (a low-pass band below 80 Hz, a band from 80 Hz to 3 kHz, and high-pass bands from 3 kHz and 9 kHz) and compressed low-level signals in each before recording, then expanded them on playback. Working band by band kept the noise reduction from audibly modulating the whole spectrum.

Broadcast processing adopted multiband designs to make stations louder and more consistent without one part of the spectrum pumping the rest. Orban’s first Optimod-FM processor, the 8000 of 1975, was a wideband design; its successor, the Optimod-FM 8100, was a two-band design. Multiband processing has been standard in broadcast chains ever since.

In music, multiband compression moved from mastering rooms into project studios in the 1990s. TC Electronic’s Finalizer, launched in 1996, packaged multiband dynamics with limiting and other mastering tools in one box. Plugins followed: Waves’ C4 became a common multiband tool, and its successor, the C6 (2010), added two floating bands and sidechain options. FabFilter Pro-MB and the dynamics modules of iZotope’s Ozone are among today’s widely used software versions.

How a multiband compressor works

Crossover filters

The crossover decides where one band ends and the next begins. The most common analog-style choice is the Linkwitz-Riley crossover, often fourth order (24 dB per octave). Its low-pass and high-pass outputs are each 6 dB down at the crossover frequency and in phase with each other, so when no compression is happening the bands add back up to a flat magnitude response. Crossover slopes vary between designs, though: some analog units use gentle 6 dB per octave crossovers instead. See crossover.

Flat magnitude doesn’t mean untouched. A Linkwitz-Riley crossover behaves like an all-pass filter: the summed output has the same frequency response as the input, but with phase shift that varies with frequency. On its own this is usually inaudible. It matters when the processed signal is blended with an unprocessed copy (as in parallel compression), where the phase difference can cause comb filtering.

Minimum phase, linear phase and dynamic phase

  • Minimum-phase (analog-style) crossovers: no added latency, phase shift around the crossover points.
  • Linear-phase crossovers: digital FIR filters that delay every frequency by the same amount, so there’s no phase shift between bands. The trade-offs are latency and pre-ringing, a faint smear before sharp transients. See linear phase EQ.
  • Hybrid modes: FabFilter Pro-MB, for example, defaults to a “Dynamic Phase” mode that Sound On Sound describes as claimed to offer the best of both worlds, with linear-phase and minimum-phase modes also available.

Per-band detection

Each band has its own detector, so a loud kick only triggers gain reduction in the low band. Many units also let a band be triggered from a different band or from an external sidechain: for example, ducking only the low end of a bass whenever the kick plays.

Time constants by band

Low frequencies have long cycles: 50 Hz takes 20 ms per cycle, 10 kHz takes 0.1 ms. Very fast attack and release on a low band can follow the waveform itself and cause distortion, so low bands usually get slower time constants than high bands. Many multiband compressors set or scale them per band for this reason.

Typical band layout

BandRough rangeWhat lives thereTypical job
LowBelow about 100 to 150 HzKick and bass fundamentals, subSteady the low end, stop it pumping the rest
Low midAbout 150 Hz to 1 kHzBody of guitars, vocals, snare; mudControl boom and boxiness when it builds up
High midAbout 1 to 6 kHzPresence, bite, sibilance zoneTame harshness on loud passages
HighAbove about 6 kHzCymbals, air, sibilanceControl splashy cymbals and esses

These ranges are starting points; good settings depend entirely on the material. Many plugins let you solo a band to hear exactly what’s in it.

Multiband compression in practice

On individual tracks

  • Vocals: a high-mid or high band acting as a de-esser, or a low-mid band catching proximity boom when the singer leans in. A static EQ cut would thin the voice all the time; the multiband only acts when the problem appears.
  • Bass: hold the low band steady while leaving the upper band, with the pick and finger noise, more dynamic.
  • Guitars: a technique popularized in metal mixing, often credited to producer Andy Sneap, uses a multiband compressor such as the Waves C4 on the rhythm-guitar bus, roughly in the 150 to 400 Hz range with fast attack and release, to take a few decibels off the low-mid buildup of palm mutes while open chords pass through.
  • Drum bus: tame cymbal wash in the top band without touching the kick and snare body.
  • Acoustic instruments: control a single resonant note or boomy range that jumps out only at certain pitches.

In mastering

Multiband compression is common in mastering, and also one of its most debated tools. Used gently, it can correct a mix whose low end swells in the chorus, or calm harsh upper mids only in the loudest sections. Used heavily, it changes the mix balance band by band, flattens the movement between sections and can make a master sound dense and lifeless. Many mastering engineers prefer to ask for a revised mix rather than fix balance problems with aggressive multiband processing, and to keep any multiband gain reduction to a few decibels at low ratios.

A careful workflow

  1. Start with all bands set to no compression, and check that the unit is transparent when bypassed and engaged (crossovers alone can change the sound slightly).
  2. Find the problem range by soloing bands, then set only that band’s threshold and ratio.
  3. Use slower attack and release on low bands, faster on high bands.
  4. Leave bands that don’t need processing alone. Using one band out of four is normal.
  5. Level-match against the bypass before deciding it’s better.

Choosing crossover points

Crossover frequencies should follow the problem, not a template. Solo the band, sweep the crossover while listening, and place it where the troublesome sound starts and stops. Keep in mind that the region around each crossover is shared by two bands, so a crossover placed right on top of an important element (the fundamental of a vocal, for example) makes that element respond to two different compressors. Wider bands with gentle settings are generally less audible than many narrow, heavily compressed ones.

Upward, downward and expansion per band

Modern multiband processors often do more than downward compression. Upward compression in a band lifts quiet detail in that range; downward expansion in a band can reduce low-level noise or room rumble only where it lives. Combining upward and downward compression in every band is how the OTT sound is made: extremely dense and bright, useful as an effect, rarely transparent.

Mid/side multiband

Some multiband processors, especially mastering suites, can process the mid (center) and side (stereo difference) signals separately in each band. That allows moves such as steadying the low end in the center while leaving the sides alone, or controlling harsh highs only at the edges of the stereo image. It’s powerful and easy to overdo; check the result in mono and in stereo.

Multiband compressor or dynamic EQ?

A dynamic EQ uses bell or shelf filters whose gain moves with the level, instead of splitting the whole signal into bands. For narrow, surgical problems (a single ringing resonance, a whistling note) a dynamic EQ is usually cleaner. For broad regions and for controlling the overall balance of the low, mid and high ranges, a multiband compressor is often the better fit. Many engineers use both.

In broadcast

Broadcast processors use multiband compression and limiting to keep stations consistently loud and balanced across very different programs. Since loudness normalization became standard in broadcasting, processing has shifted toward meeting loudness targets with fewer artifacts rather than simply maximizing level.

Multiband compressors: gear and software

Some links in this section go to Thomann and are affiliate links: if you buy through them, musicproductionglossary.com may earn a small commission at no extra cost to you. Learn more.

  • Drawmer 1973: a three-band stereo FET compressor with adjustable crossovers and a Mix control; a popular hardware choice for mix bus and mastering work.
  • Tube-Tech SMC 2B: a high-end stereo multiband master compressor with three compressor bands per channel.
  • FabFilter Pro-MB: a plugin with up to six freely placed bands and Dynamic Phase, linear-phase and minimum-phase processing.
  • Waves C6: four crossover bands plus two floating bands, with sidechain options.
  • Waves C4: the four-band plugin behind many classic mixing techniques.
  • iZotope Ozone 12 Standard and Ozone 12 Advanced: mastering suites that include a multiband Dynamics module.

Most DAWs also include a multiband compressor: Ableton Live’s Multiband Dynamics (where the OTT preset started), Logic Pro’s Multipressor and others.

Common mistakes and myths

“A multiband compressor is just a fancy EQ”

An EQ applies the same change all the time. A multiband compressor changes each band only when its level crosses the threshold, so it can fix problems that come and go.

Compressing every band by default

Turning all bands on with similar settings tends to flatten the mix and shift its tonal balance. Target the band that needs it and leave the others alone.

“Linear phase is always better”

Linear-phase crossovers avoid phase shift but add latency and pre-ringing, which can soften sharp transients. Minimum-phase crossovers are often the better choice on percussive material. Use linear phase when phase coherence with another signal really matters.

Fast time constants on the low band

Very fast attack and release on low frequencies cause distortion. Give the low band slower settings.

Fixing the mix in mastering

If a mix needs heavy multiband compression to sound balanced, the problem is usually in the mix. Fixing it there is often more effective.

ProcessorHow it splits the signalWhat it changesBest for
Multiband compressorCrossovers divide the whole signal into bandsLevel of each band, independentlyBroad frequency-specific dynamics, mastering
Single-band compressorNo splitLevel of the whole signalOverall leveling and glue
Dynamic EQNo split; individual EQ filtersGain of one EQ band, only when neededNarrow, surgical problems
De-esserSidechain filter, sometimes a split bandSibilance range or whole signalEsses on vocals
Static EQFiltersFixed tonal balancePermanent tone shaping
OTTThree bandsUpward and downward compression per bandExtreme density as an effect

Standards and references

The crossover theory used in multiband processors goes back to Siegfried Linkwitz’s work on active crossovers for loudspeakers, published by the Audio Engineering Society in the 1970s, and Rane’s technical notes remain a widely cited practical reference. On the broadcast side, loudness standards such as ITU-R BS.1770 and EBU R 128 changed how multiband processors are used, because loudness is now measured and normalized rather than simply maximized.

  • Compressor: the single-band version.
  • Crossover: the filters that split the bands.
  • Linear phase EQ: the same linear-phase trade-offs apply to crossovers.
  • De-esser: a specialized frequency-dependent dynamics processor.
  • OTT: an extreme multiband setting.
  • Sidechain: per-band and external triggering.
  • Mastering: where multiband compression is most debated.
  • EQ: static tone shaping, the usual first tool.

FAQ

What does a multiband compressor do?

It splits the audio into frequency bands, compresses each band separately, then recombines them, so a loud low end doesn’t turn down the highs and vice versa.

What is the difference between a multiband and a regular compressor?

A regular compressor turns the whole signal down when it gets loud. A multiband compressor only turns down the band that got loud.

When should I use a multiband compressor?

When a problem in one frequency range comes and goes: boomy notes, harshness on loud passages, esses, a low end that swells in the chorus.

Is a multiband compressor the same as dynamic EQ?

No. A multiband compressor splits the whole signal with crossovers; a dynamic EQ applies level-dependent gain to individual EQ bands. Dynamic EQ is often cleaner for narrow problems.

Should I use a multiband compressor on the master?

With care. Light settings can correct specific problems; heavy settings change the mix balance and flatten the dynamics.

How many bands do I need?

Three or four is typical. Often only one band actually needs compression.

Why does my multiband compressor change the sound even with no compression?

The crossover filters themselves can shift phase or, in linear-phase mode, add pre-ringing. Check how transparent the unit is with all bands set to no compression.

Do I need linear phase mode?

Mainly when the result is blended with an unprocessed copy or when phase coherence is critical. Otherwise minimum-phase crossovers avoid latency and pre-ringing.

How should I set attack and release per band?

Slower on low bands to avoid distortion, faster on high bands to catch quick peaks like cymbals and esses.

Can a multiband compressor work as a de-esser?

Yes. Compressing only a high-mid or high band set around the sibilance range works like a split-band de-esser.

What is OTT?

An extreme three-band upward and downward compression setting that began as a preset in Ableton Live’s Multiband Dynamics and became a free Xfer Records plugin.

Who invented multiband compression?

Band-split dynamics processing first appeared in Dolby’s A-type noise reduction (1965). Broadcast processors and later mastering tools such as TC Electronic’s Finalizer brought multiband compression into common use.