Multiband Compression
Multiband compression splits an audio signal into two or more discrete frequency bands using crossover filters, then applies independent compression to each band before recombining them at the output. Unlike full-band compression, each band has its own threshold, ratio, attack, release, and makeup gain controls, allowing dynamic control that responds differently across the frequency spectrum. The result is a tool capable of taming low-end boom without touching high-frequency air, controlling muddy mids without squashing transient snap, and evening out tonal balance without the pumping artifacts of a single-band compressor.
Multiband compression makes mixes louder and more commercial, so more bands and more gain reduction equals a better, more professional master.
Multiband compression does not inherently add loudness — it manages frequency-domain dynamics to prevent any one band from triggering limiting that would otherwise reduce overall perceived loudness. Over-applying it creates phase artifacts, tonal hollowness, and dynamic lifelessness that makes a master sound worse than a well-executed single-band approach. Professional mastering engineers routinely achieve louder, more dynamic masters with minimal multiband processing than amateur applications with maximal band-splitting.
What Is Multiband Compression?
When one compressor compresses everything equally, it compresses nothing intelligently — multiband is the art of teaching dynamics to respect frequency.Multiband compression splits an audio signal into two or more discrete frequency bands using crossover filters, then applies independent compression to each band before recombining them at the output. The critical distinction from single-band compression is that each band operates in its own dynamic bubble — a boom in the sub doesn't trigger gain reduction in the upper mids, and a harsh vocal consonant doesn't pull the low end down with it. You're not compressing a signal; you're compressing a collection of frequency-specific signals that happen to share the same audio stream.
The practical consequence of this architecture is immediate and profound. When a kick drum hits on a full-band compressor, the entire mix ducks — the bass, the mids, the air, everything responds to the transient. With multiband, you can isolate that low-band energy and compress it independently, leaving the rest of the spectrum untouched. The result is a mix that breathes naturally in the upper registers while remaining controlled and tight below 100 Hz. This is why multiband compression became the dominant tool in professional mastering and mix bus processing — it solves the core contradiction of full-band dynamics control, which is that the tool you use to fix one frequency range inevitably damages another.
Each band in a multiband compressor carries its own threshold, ratio, attack, release, and makeup gain — the same parameters you'd find on any compressor, replicated independently per band. The crossover points that define where one band ends and the next begins are themselves adjustable, which means you're not just setting compression parameters but actively sculpting the frequency topology of your dynamic control. Moving a crossover from 200 Hz to 300 Hz on a mix with a congested low-mid region can be as audibly significant as adjusting EQ. Every crossover position is a decision with consequences, which is why multiband rewards deliberate use and punishes casual deployment.
This tool is simultaneously the most powerful and most misused processor in a modern production chain. Producers who treat it like a glorified EQ — cranking ratios band by band until something sounds louder — routinely create masters that are dynamically dead, tonally inconsistent across playback systems, and fatiguing at full volume. The engineers who use it correctly apply 2–4 dB of gain reduction per band, at slow attack settings that let transients breathe through before compression engages, and leave two or three bands entirely inactive unless there's a specific, identifiable problem in that range. The discipline is in restraint, not reach.
Understanding multiband compression also requires understanding what it cannot do. It doesn't fix a poorly recorded source — a bass with uneven fret noise across the neck will still have uneven fret noise after multiband processing, just with different gain levels applied to specific frequency ranges. It doesn't replace targeted subtractive parametric EQ for surgical frequency correction. And it doesn't substitute for proper gain staging — a mix that's peaking wildly before the multiband insert will produce gain reduction artifacts that no crossover adjustment can recover. The tool works best on a signal that's already 80% correct, where its job is refinement and balance rather than rescue.
— Soulwax (David & Stephen Dewaele), Producers/Artists — Sound On Sound — Soulwax: The Production Philosophy, February 2018"Multiband compression on a mix bus is a surgical tool. You compress the low end differently from the mids differently from the highs. The result is a master that feels balanced at any volume."
Multiband compression divides the signal into frequency-specific bands and applies independent dynamic control to each, solving problems that a single-band compressor would create by treating the entire spectrum as a single entity.
How Multiband Compression Works
The signal flow begins at a bank of crossover filters — typically Linkwitz-Riley or Butterworth designs — that divide the incoming audio into non-overlapping frequency ranges. A four-band implementation might route everything below 80 Hz to Band 1, 80–500 Hz to Band 2, 500 Hz–8 kHz to Band 3, and above 8 kHz to Band 4. These filters are linear-phase or minimum-phase depending on the design, and that distinction matters: minimum-phase crossovers introduce phase shift at the crossover frequencies, which can cause subtle comb-filtering artifacts when bands are recombined. Linear-phase designs eliminate this problem but introduce pre-ringing and latency. Neither is universally better — they're different tools for different contexts, and every professional multiband plugin makes a deliberate choice between them.
Each filtered band feeds into its own dedicated compressor circuit that responds only to the energy within that band's frequency range. This is the sidechain principle at the core of multiband operation: the detector in each band reads only that band's level to determine how much gain reduction to apply. When a 60 Hz sub hit pushes Band 1 above threshold, the gain reduction in Band 1 doesn't know or care that Band 3 is sitting quietly — Band 3 passes through unaffected. The compressor in each band is functionally identical to a standard compressor in terms of its parameters and behavior; the only difference is that its input and sidechain are both band-limited. Some multiband designs allow external sidechain signals per band, which enables frequency-selective ducking — a technique used extensively in broadcast and in modern pop production to duck specific frequency ranges of music beneath dialog or vocals.
After gain reduction is applied independently in each band, the processed bands are summed back together at the output stage. This recombination is where crossover phase characteristics become audible — if the filters introduce phase discrepancies between bands, the summed output will have frequency response anomalies at the crossover frequencies even when no compression is active. Well-designed multiband processors ensure that the summed output of all bands at unity gain and zero compression is phase-coherent and sonically transparent, meaning the processor should be inaudible when bypassed at matched gain. Testing any multiband compressor by bypassing it at matched levels is the first diagnostic step before committing to settings — if you hear a tonal change with compression off, the crossover design is coloring your signal before compression even enters the picture, and that characteristic needs to be factored into every decision you make with that unit.
Crossover filters route specific frequency ranges to dedicated compressor circuits that act only on energy within that band, then the processed bands are phase-coherently summed at the output.
Multiband Compression — Key Parameters
Every multiband compressor multiplies these parameters by the number of active bands, which means a four-band unit has four thresholds, four ratios, four attack settings, four release settings, four makeup gains, and three crossover frequency points — sixteen or more simultaneous decisions, all of which interact with each other and with the source material. The discipline of multiband compression is knowing which parameters to touch and which to leave at default, because changing all of them simultaneously produces confusion rather than control.
The crossover point is the most consequential parameter in the entire unit — move it wrong and you're compressing the wrong content. Set the low-band crossover at 80 Hz for sub-heavy music and watch the kick drum become the sole trigger for low-band compression. Move it to 120 Hz and the bass guitar joins the party. That 40 Hz shift completely changes which elements drive the gain reduction in your most sensitive band. Sweep crossover points slowly while watching the gain reduction meters to identify where compression starts responding to the right content.
Set the threshold where the problem actually lives, not where the signal averages. Low-band thresholds often sit lower than mid-band because sub content arrives with high peak energy from kick and bass. A low-band threshold at -18 dBFS will compress on every kick hit; at -12 dBFS it only engages on the loudest hits. On a mix bus, 2–4 dB of gain reduction per band is the professional target — if you're hitting more than 6 dB reduction in any band, either your source has a real imbalance problem or your threshold is set too low.
In multiband mastering, anything above 4:1 in a single band starts sounding like that frequency range is being clamped rather than controlled. A 2:1 ratio across three bands with consistent gain reduction produces a more natural-sounding master than a 6:1 ratio on one problem band. Limiting ratios (10:1 and above) in a multiband context are appropriate for broadcast loudness maximization but will make individual frequency ranges feel dead at high gain reduction amounts — the low band stops pumping, the high band stops breathing.
Attack behavior differs meaningfully across frequency bands because transient content is not evenly distributed across the spectrum. Low frequencies have longer wavelengths, so the compressor needs a slower attack — 20–40ms — to respond to the actual peak energy rather than the slope of the waveform. High-frequency transients are short and sharp; a 5ms attack on the high band will catch consonant spikes cleanly. Set attack too fast in the low band and kick transients are clipped before the compressor can distinguish them from sustained bass — the punch disappears and the kick sounds smeared.
Release sets how quickly each band returns to unity after gain reduction. Too fast in the low band and you get pumping — the sub range surges between kick hits, creating an audible breathing artifact that becomes unbearable on headphones at high volume. Too slow and the compression doesn't recover before the next transient, causing progressive gain reduction that flattens the low band entirely over time. Auto-release modes in modern plugins analyze the signal's envelope and adjust continuously — on complex program material like a full mix, this outperforms any fixed release time you can dial in manually.
Makeup gain in multiband is a tonal equalizer disguised as a level control. Adding 2 dB of makeup to the high band after compression makes the top end brighter — identical to boosting a high shelf with a shelving EQ. This means every makeup gain decision is simultaneously a compression decision and an EQ decision, which is why multiband processing is so powerful for mastering — you're shaping dynamics and tonal balance simultaneously with a single processor. Always use an output meter and null test to confirm makeup gain is matching levels, not boosting them.
The interaction between attack and crossover frequency is the most underestimated relationship in multiband compression. A crossover set at 200 Hz that also catches the body of a snare drum — common in mixing contexts — means the mid-band attack setting now directly shapes snare transient character. Set that attack at 8ms and snare hits compress before the crack clears the detector; set it at 25ms and the entire transient passes through before gain reduction kicks in. The same attack setting produces completely different sonic results depending solely on where the crossover sits relative to the source's harmonic content. This is why crossover frequency and attack should always be adjusted together, not in isolation.
Gain reduction metering across all bands simultaneously is the professional's primary diagnostic tool. On a healthy mix bus with appropriate multiband settings, all three or four meters should be moving — but none should be pinned. If Band 2 (low-mid) is showing 8–10 dB of gain reduction while Bands 1, 3, and 4 are barely moving, the source has a low-mid buildup problem that multiband compression is working hard to contain. The correct response is to first address the problem with subtractive EQ before the multiband insert, then dial back the Band 2 threshold so the compressor is handling the residual variation rather than the bulk of the tonal imbalance. Multiband compression is not a substitute for mix bus EQ — it works in concert with it.
Each band's threshold, ratio, attack, and release interact with the crossover frequency to determine which source content triggers compression — making crossover placement the most consequential parameter in the entire unit.
Quick Reference Card
In professional mastering contexts, limiting each band's gain reduction to a maximum of 3 dB preserves the dynamic integrity and tonal balance of the mix while achieving meaningful frequency-domain control. Exceeding 3 dB per band consistently begins to introduce audible coloration, hollowness, or pumping that degrades rather than enhances the master.
These starting points assume a properly gain-staged signal at the multiband input — adjust thresholds first, then ratio, then attack and release to taste.
| Source | Ratio | Attack | Release | Threshold | Notes |
|---|---|---|---|---|---|
| Mix Bus (Mastering) | 2:1 per band | 30ms low, 15ms mid, 8ms high | Auto or 150ms | -18 to -12 dBFS | 2–4 dB max GR per band; linear-phase crossovers preferred |
| Bass Guitar | 3:1 low band, 2:1 mid band | 20ms low, 10ms mid | 100ms low, 80ms mid | -20 dBFS low | Crossover at 80 Hz and 250 Hz; tighten sub without dulling finger noise |
| Drum Bus | 2.5:1 low, 2:1 mid | 40ms low, 20ms mid, 5ms high | 200ms low, 100ms mid | -16 dBFS | Let kick transient through low band; use high band for cymbal control only |
| Vocal (Pop) | 2:1 low-mid, 3:1 high | 15ms low-mid, 5ms high | 80ms | -14 dBFS | Low-mid band targets boxy buildup at 300–500 Hz; high band tames sibilance |
| Full Mix (EDM Master) | 3:1 low, 2:1 mid, 2:1 high | 10ms low, 15ms mid, 5ms high | Auto all bands | -12 dBFS low, -16 mid/high | Sub band critical; crossover at 80 Hz must isolate kick and bass cleanly |
| Acoustic Guitar | 2:1 low-mid, 2:1 high | 25ms low-mid, 10ms high | 120ms | -18 dBFS | Low-mid band targets body resonances 200–400 Hz; high band controls pick attack |
| Podcast / Broadcast VO | 4:1 low, 3:1 mid | 10ms | 60ms | -20 dBFS low, -16 mid | Low band removes proximity effect; mid band controls proximity-driven nasal peaks |
Tools for This Entry
Signal Chain Position
Multiband compression sits after single-band compression and after primary EQ shaping, but before a final limiting stage. The logic is sequential: single-band compression handles overall dynamic range and glue, EQ corrects tonal balance problems, and then multiband compression addresses the residual frequency-specific dynamic inconsistencies that remain. Inserting multiband before EQ means you're compressing the problem frequencies rather than removing them — the multiband will work harder and produce more audible artifacts. On a mastering chain specifically, multiband typically sits between a bus compression stage and a final true peak limiting stage, with its makeup gains calibrated so the limiter receives a consistently leveled, tonally balanced signal that can be pushed to the output ceiling without spectral tipping.
Interaction Warnings
- Crossover Phase vs. Stereo Width: Minimum-phase crossover designs introduce frequency-dependent phase shift that collapses differently in left and right channels under heavy compression, narrowing stereo imaging at crossover frequencies. On a stereo mix bus, always prefer linear-phase designs and verify stereo width with a correlation meter before committing to a multiband preset.
- Mid-Side Interaction: Many mastering-grade multiband compressors operate in mid-side processing mode, compressing the mono center and stereo sides independently per band. This is extraordinarily powerful but doubles the parameter count — a four-band M/S unit has eight independent compressors running simultaneously. A common error is over-compressing the side channel's low band, which eliminates bass width that was intentional in the mix.
- Limiting Interaction: A multiband compressor preceding a limiting stage can create inconsistent limiter behavior if band makeup gains are misaligned. If the low band's makeup gain is +3 dB and the high band's is +1 dB, the limiter receives a low-heavy signal and will clamp the low band disproportionately on peaks, reintroducing the spectral imbalance you just corrected.
History of Multiband Compression
The functional precursor to multiband compression appeared in broadcast engineering during the 1950s and 1960s, where AM radio transmitters required that different frequency ranges be controlled independently to prevent over-modulation on bass-heavy content without sacrificing high-frequency presence. Early broadcast processors used passive filter networks and separate limiting amplifiers for each band — crude by modern standards but solving the same fundamental problem: low-end peaks were destroying loudness headroom that the rest of the spectrum wasn't using. These systems weren't musical tools; they were technical compliance equipment. The creative application of frequency-selective dynamic control came later, when studio engineers began noticing that what worked for broadcast modulation also worked for tonal consistency in music production.
The hardware golden age of multiband compression arrived in the 1970s and 1980s through units like the Orban Optimod series — broadcast processors that became studio staples for their ability to control low-end energy independently of mid and high-frequency content. The dbx 3BX Dynamic Range Enhancer took the concept into hi-fi and studio applications, offering three-band expansion that producers quickly reversed for compression purposes. By the late 1980s, the Neve 33609's program-dependent behavior and the Focusrite Red series inspired thinking about frequency-selective processing, while dedicated mastering engineers like Bob Ludwig and Doug Sax were using crossover-based processing in mastering chains that they guarded closely as trade secrets. These engineers understood that the difference between a good and great master often came down to which frequency bands were being asked to do what work dynamically.
The software era transformed multiband compression from an expensive hardware specialty into a democratized production tool. Waves introduced the C4 Multiband Compressor in the late 1990s — arguably the plugin that put the technology in front of every DAW user worldwide. McDSP's ML4000 followed, offering mastering-grade precision at plugin prices. TC Electronic's MD3, Sonnox Oxford Dynamics, and later iZotope Ozone's multiband module each added refinements: linear-phase crossovers, dynamic EQ hybrid modes, M/S processing per band, and auto-gain compensation. The precision available in software exceeded what hardware units could offer — digital crossovers could be set to exact frequencies with no component drift, and parameter automation became possible for the first time. What was lost was the character of analog circuitry in the gain reduction stage — software compressors compressed accurately but not always musically, which drove demand for hardware-emulation plugins that reintroduced transformer saturation and VCA coloration into the digital processing chain.
In the streaming era, multiband compression's role has been reshaped by loudness normalization and LUFS targeting. The arms race of the 2000s — where multiband limiting was weaponized to push integrated loudness beyond -8 LUFS — became largely irrelevant when Spotify, Apple Music, and YouTube all adopted normalization targets around -14 LUFS integrated. Masters that were crushed to -6 LUFS now get turned down on streaming platforms, and their dynamic range damage is exposed rather than rewarded. Contemporary mastering engineers use multiband compression with considerably more restraint than a decade ago, targeting tonal consistency and translation across playback systems rather than raw loudness maximization. The tool's power remains unchanged; the incentive to abuse it has been structurally removed by platform normalization algorithms.
— Bob Clearmountain, Mix Engineer (Bruce Springsteen, The Rolling Stones, Bryan Adams) — Tape Op Magazine Issue 78, 2010"I use compression to make things feel right, not to make them louder. There's a difference and most people confuse the two."
Multiband compression evolved from broadcast over-modulation control through hardware mastering specialty to democratized DAW tool, and its contemporary use is defined by streaming platform normalization targets that reward dynamic integrity over raw loudness.
How Producers Use Multiband Compression
The professional workflow starts with identification, not application. Before inserting a multiband compressor, use a spectrum analyzer to identify which frequency ranges are dynamically inconsistent — not just which are loudest, but which are most variable in level over time. A bass line that swings 8 dB between root notes and octave positions is a low-band problem. A vocal that sounds boxy on chest voice but thin on head voice is a low-mid problem. A mix with inconsistent brightness between choruses and verses is a high-band problem. Once you've identified the specific frequency ranges that need independent dynamic control, insert the multiband compressor and configure it with only those bands active, leaving all other bands in bypass or at unity. This targeted approach produces transparent results because inactive bands pass through with zero processing artifacts.
For mix bus applications, the standard professional workflow is to set crossovers first, verify transparency with no compression active, then bring in gain reduction one band at a time starting from the lowest band. Watch the gain reduction meter for each band and target 2–3 dB of reduction on average, with peak reduction no higher than 6 dB in any single band. If a band requires more than 6 dB to sound controlled, the source has a structural problem that multiband is being asked to compensate for — the correct response is to return to the mix, fix the imbalance in the individual tracks, and then reapply multiband compression to what remains. Use parallel compression on the multiband output when you need to preserve the attack character of the uncompressed signal while still benefiting from the spectral balance correction — blend 30–40% compressed signal with 60–70% dry signal for a mix bus that feels controlled but alive.
1. Insert Ableton's native Multiband Dynamics device on your track or mix bus (found under Audio Effects > Dynamics). 2. The device defaults to three bands; click the crossover frequency handles in the display to adjust split points — drag the low/mid crossover to approximately 120 Hz and the mid/high crossover to approximately 4 kHz. 3. Click each band's header to activate it and set independent Above (compressor) and Below (expander) thresholds by dragging the colored handles in the display. 4. Set the Ratio for each band using the Above Ratio knob — start at 2:1 for mastering or 3:1 for mixing. 5. Adjust Time (combined attack/release) per band or switch to manual control for individual attack and release settings. 6. Use the Output knob per band for makeup gain. 7. Enable the Soft Knee option for more transparent processing. 8. Use the global In and Out meters to verify gain-matched A/B bypass comparison.
1. Insert Logic Pro's native Multipressor (found under Dynamics in the plugin browser) on your channel strip. 2. The Multipressor shows four bands by default with a frequency analyzer display — click and drag the crossover triangles to set split points at 120 Hz, 500 Hz, and 4 kHz as a starting point. 3. Click each band to select it, then set Threshold, Ratio, Attack, and Release in the parameter section below the display. 4. The Compression Amount bar shows real-time gain reduction per band — aim for 2–4 dB maximum during peaks. 5. Set Output Gain per band as makeup gain. 6. Enable the Lookahead parameter (under the master settings) for mastering contexts. 7. Use the Bypass button per band to A/B individual bands. 8. The Spectral Display option in the settings enables a real-time per-band frequency overlay for visual feedback.
1. Insert Maximus from the native plugin browser (under Dynamics) on your mixer track. 2. Maximus opens with three bands (Low, Mid, High) visible in the top band-assignment display — click and drag the crossover sliders to adjust split frequencies; try 120 Hz for low/mid and 4 kHz for mid/high. 3. Click each band tab (Low, Mid, High, or Master) to access its individual compressor parameters: Threshold, Ratio, Attack, Release, Sustain, and Stereo Separation. 4. The tension-graph displays show the compressor curve for each band — adjust by clicking and dragging the nodes. 5. Use the Post-Gain knob per band for makeup gain. 6. Enable the LMH Pre-clip option to prevent inter-band clipping artifacts. 7. The Visualizer tab shows real-time gain reduction per band overlaid on a frequency display. 8. Use the Master tab to apply overall output limiting and final ceiling control after all band processing.
1. Insert a third-party multiband plugin (Waves C6, FabFilter Pro-MB, or iZotope Ozone Dynamics) on your Aux, Master Fader, or track insert — Pro Tools has no native multiband compressor. 2. For Waves C6: set the six crossover points across the frequency display by clicking and dragging the band boundaries — start with bands centered at 80 Hz, 240 Hz, 750 Hz, 2.4 kHz, and 6 kHz. 3. Set Threshold, Ratio, Attack, and Release per band using the individual channel strip controls. 4. Activate Makeup Gain per band or use the Master Output for overall level compensation. 5. Use the Couple option to link adjacent bands for more natural compression interaction. 6. For FabFilter Pro-MB: drag the band handles directly on the spectrum analyzer display, set each band to Dynamic mode, and use the Range parameter to limit maximum gain reduction. 7. Enable plugin delay compensation in Pro Tools (Setup > Playback Engine > check 'Delay Compensation') to prevent phase offset from linear-phase plugins. 8. Compare A/B using Pro Tools' plugin bypass shortcut (Command+click the insert) with session volume matched at the mix bus output.
The listening test for multiband compression is frequency-specific bypass. Don't bypass the entire unit — bypass individual bands while music plays and listen to what changes. If bypassing the low band makes the mix feel looser and more unruly in the sub range, the band is doing real work. If bypassing it makes no audible difference, you either don't need that band active or the threshold is set too high to engage on real program content. The goal is that every active band should produce an audible difference when bypassed — but the processed version should sound more controlled, not louder. If the compressed version sounds louder on bypass comparison at matched output levels, you've added makeup gain without adequate compression, and the perceived improvement is just volume illusion.
In production contexts outside of mastering — on individual tracks, buses, and stems — multiband compression often works best combined with automation rather than replacing it. A bass guitar with three problem notes across the neck is better served by clip gain correction on those three notes followed by gentle multiband compression than by heavy multiband settings across the entire performance. Automation handles deterministic, repeatable problems; multiband compression handles stochastic, unpredictable dynamic variation. Using both together means the multiband's gain reduction meter is moving gently and consistently throughout the track rather than spiking violently on the three notes you should have fixed in the edit.
Identify which frequency ranges are dynamically inconsistent before inserting the compressor, apply gain reduction one band at a time, and use per-band bypass to verify every active band is doing audible and necessary work.
Multiband Compression by Genre
Multiband compression settings vary dramatically across genres because each genre has fundamentally different energy distribution across the frequency spectrum — hip-hop's sub-heavy low end demands different low-band thresholds than folk music's midrange-dominant acoustic content, and EDM's maximized loudness targets require different ratio and release behaviors than jazz's preserved dynamic range.
| Genre | Ratio | Attack | Release | Threshold | Notes |
|---|---|---|---|---|---|
| Trap | 4:1–8:1 | 5–20ms | 80–150ms | -18 to -24 dBFS | Heavy low-band compression to control 808 sub bloom; high band kept light to preserve hi-hat crispness and brightness |
| Hip-Hop | 3:1–5:1 | 10–30ms | 100–200ms | -12 to -18 dBFS | Low-mid band compression (200–600 Hz) to control sample muddiness; gentle high-band to preserve vocal presence and clarity |
| House | 3:1–6:1 | 5–15ms | 150–300ms | -14 to -20 dBFS | Sub band set to maintain consistent four-on-the-floor kick weight across the full track; mid band gentle to keep synth pads open |
| Rock | 2:1–4:1 | 15–40ms | 100–250ms | -10 to -16 dBFS | Low-mid band tames 200–400 Hz guitar/kick competition; slower attacks preserve transient snap of snare and guitar pick attack |
| Mastering | 1.5:1–3:1 | 30–80ms | 200–500ms | -6 to -12 dBFS | Maximum 3 dB gain reduction per band; linear-phase mode preferred; focus on tonal balance correction rather than loudness maximization |
When a track crosses genre boundaries — a pop song with trap production elements, a rock record with EDM-influenced mastering — default genre settings become less useful than source-based analysis. Trust the spectrum analyzer and the gain reduction meters over any genre template, and adjust crossover frequencies to match the actual harmonic content of your specific mix rather than the genre's statistical average.
Hardware vs Plugin vs Stock
The meaningful differences between hardware and plugin multiband compressors are crossover quality, gain reduction character, and workflow ergonomics — not raw processing power. Hardware units like the Weiss DS1-MK3 and SPL IRON compress with analog gain reduction circuits that introduce specific harmonic coloration at high gain reduction amounts, a character that digital plugin designs must actively emulate rather than inherently possess. Plugin multiband compressors have the advantage of precise parameter recall, LUFS metering integration, linear-phase crossover options without additional latency penalty in offline processing contexts, and automation capabilities that hardware cannot match. Stock DAW multiband compressors — Logic's Multipressor, Ableton's Multiband Dynamics — are functionally competent for mixing applications but typically lack the metering resolution and crossover precision of dedicated mastering-grade plugins.
| Aspect | Hardware | Plugin |
|---|---|---|
| Crossover Quality | Analog filter design; inherent phase characteristics; component-dependent | Digital precision; choice of linear-phase or minimum-phase; no drift |
| Gain Reduction Character | VCA, optical, or transformer-coupled coloration at high GR amounts | Clean by default; emulation plugins add modeled coloration |
| Parameter Recall | Manual recall via session notes or recall sheets; no automation | Instant recall; full parameter automation in DAW timeline |
| Metering | Analog VU or discrete LED meters; limited resolution | High-resolution digital metering; per-band LUFS display in modern plugins |
| Latency | Near-zero inherent latency; suitable for live processing | Linear-phase designs introduce look-ahead latency; compensated automatically in DAW |
| Cost | $2,000–$15,000 for mastering-grade hardware | $50–$500 for professional-grade plugins; stock options free with DAW |
Before and After
The mix has an inconsistent low end that blooms on kick hits and muddies the bass line, the low-mids feel congested and boxy around 300–400 Hz, and the top end occasionally gets harsh during dense sections — the mix sounds different on headphones versus speakers, with the low end either too dominant or too thin depending on playback system.
The low end sits locked and consistent — the sub energy is present and controlled whether the kick hits or not, the bass line maintains its weight without blooming. The low-mid boxiness is tamed and the body of the mix feels full but not congested. The top end maintains air and presence even during the densest moments without harshness, and the mix translates with remarkable consistency across earbuds, studio monitors, and club systems.
When listening to the before-and-after comparison, resist evaluating overall loudness — match output levels and listen specifically for frequency-range behavior. In the before state, a kick drum should cause the entire mix to duck slightly as a full-band compressor responds to the transient; in the after state with multiband engaged, only the low band should respond to the kick, leaving mids and highs unaffected. A well-tuned multiband setting makes the after version feel more spacious and consistent simultaneously — tighter in the low end, cleaner in the mids, without any single frequency range dominating the dynamic behavior of the whole mix.
Multiband Compression In The Wild
These tracks demonstrate multiband compression not as an effect but as an architectural decision — each one represents a case where frequency-specific dynamic control is the reason the mix holds together at high volume without sounding crowded, or the reason the low end sits controlled without dulling the top. Listen actively with each example, targeting the specific frequency range mentioned rather than the overall sound.
The consistent lesson across all seven tracks is that effective multiband compression is invisible — you hear the result (a controlled, balanced, full-sounding mix) without hearing the process. When multiband compression is audible as a processing artifact (pumping in a specific frequency range, tonal tipping as dynamics shift, narrowed stereo width on loud sections), it has been over-applied or misconfigured. The goal in every one of these productions was for listeners to feel the music, not notice the compression.
Types of Multiband Compression
See the full comparison: Parallel Compression
See the full comparison: Limiting
Multiband compression is not a single technology but a family of related approaches, each optimized for different use cases. The differences between types are not cosmetic — choosing the wrong type for your application can introduce phase artifacts, destroy transient character, or create loudness-dependent tonal shifts that undermine the entire point of frequency-selective processing. Know which type you're using before you dial in a single parameter.
Linear-phase crossovers maintain phase coherence across all frequency bands, eliminating comb-filtering artifacts at crossover frequencies. Essential for mastering and mix bus processing where phase integrity is non-negotiable. The tradeoff is look-ahead latency and pre-ringing on transients — on program material with sharp transients like drums, this can create subtle smearing before the transient that is inaudible in context but shows on a waveform zoom.
Minimum-phase designs introduce frequency-dependent phase shift at crossover points but have zero latency and no pre-ringing. Better suited for live processing, mixing applications where real-time monitoring is required, and sources where slight phase shift at crossover frequencies is acceptable or even musically useful. The coloration from minimum-phase crossovers can add a subtle density at crossover frequencies that some engineers prefer on individual tracks.
Dynamic EQ applies frequency-specific gain reduction using parametric EQ band shapes rather than crossover-defined regions, meaning each band can be a bell, shelf, or notch filter with adjustable Q. This produces more surgical control than traditional multiband crossovers and is ideal for targeting specific resonant problems — a 3 dB bass boom at 80 Hz, a harsh 3 kHz peak in vocals — without affecting adjacent frequencies. The distinction from traditional multiband compression is that the bands can overlap, and each band's shape is adjustable independent of its dynamic behavior.
M/S multiband applies independent compression to the mono center (mid) and stereo sides (side) in each frequency band simultaneously — effectively running eight or more compressors on a stereo signal. This unlocks the ability to tighten the sub range only in the center while leaving low-mid width in the sides untouched, or to control vocal presence in the mid channel without affecting the stereo reverb returns in the side. Reserved for mastering and advanced mix bus applications where stereo field control per frequency band is a deliberate creative decision.
Designed for maximum loudness within a target integrated level, these units use high-ratio limiting per band combined with aggressive lookahead to push each frequency range to its individual ceiling simultaneously. The result is maximum perceived loudness at the expense of dynamic integrity — appropriate for broadcast compliance and streaming loudness maximization where LUFS targets are mandatory, but inappropriate for music mastering where dynamic range preservation is a sonic priority.
Rather than reducing gain above threshold, upward compression raises gain below threshold per band — expanding quiet passages in specific frequency ranges rather than compressing loud ones. Used to restore low-end body that was lost during tracking, to add presence to a thin-sounding vocal in the upper midrange without boosting louder passages, or to enhance headroom by lifting quiet sections band-selectively rather than applying makeup gain globally.
Choosing the correct type of multiband compression — linear-phase vs. minimum-phase, traditional crossover vs. dynamic EQ, downward vs. upward — is as important as any parameter setting, because each type introduces different artifacts and enables different degrees of tonal precision.
The single biggest multiband compression mistake is using it before you've diagnosed what's actually wrong. Producers reach for multiband the same way beginners reach for compression — as a first-pass fix for everything — and end up with mixes that feel simultaneously over-processed and dynamically inconsistent. Multiband earns its place in a chain only when you can name the specific frequency range that needs its own compression behavior and explain exactly why the single-band compressor before it can't solve the problem. If you can't answer that question in one sentence, you don't need multiband yet. If you can, it's the most precise tool in the building.
Every band you leave inactive is a decision that makes you a better engineer — restraint is the technique.
Common Mistakes with Multiband Compression
Multiband compression is uniquely capable of creating problems that sound like solutions — a mix can feel more controlled, more present, and louder after heavy multiband processing, while simultaneously losing the dynamic integrity that made it feel musical in the first place. The mistakes below are not beginner errors. Professional engineers with years of experience make several of them regularly, especially under deadline pressure when the temptation to reach for a preset outweighs the patience to diagnose before compressing.
Loading a preset and trusting the default crossover positions without verifying they match your source material is the fastest route to compressing the wrong content. A default low/mid crossover at 200 Hz on a hip-hop track where the kick's fundamental sits at 60 Hz and the bass line occupies 80–180 Hz means your low band is compressing both sub content and bass body simultaneously. Move that crossover to 100 Hz and suddenly you can control the sub independently of the bass's midrange harmonics. Always sweep crossover frequencies while watching gain reduction meters to confirm each band is responding to the intended content.
When a frequency range needs constant, heavy gain reduction — 8–10 dB consistently throughout the track — that's a level problem, not a dynamic problem. Using multiband compression to reduce a permanently loud low-mid range is equivalent to using a compressor instead of a fader. The correct solution is subtractive EQ to remove the excess energy, then gentle multiband compression to handle the residual dynamic variation. Multiband set to constant heavy reduction is an expensive way to do what a 3 dB EQ cut could accomplish transparently.
Using the same attack time across all bands ignores the fundamental physics of frequency and transient duration. Low-frequency transients have longer wavelengths and require slower attack times — a 5ms attack on a sub band will clip the front of a kick drum's fundamental before the detector can resolve it. High-frequency transients are brief and sharp; a 40ms attack on the high band lets every sibilant consonant and cymbal splash pass through uncompressed. Match attack time to the transient content within each band's frequency range, not to a single value applied globally.
The low band is the most sensitive band in any multiband configuration because the human ear is most attuned to sub-frequency dynamic variation — pumping in the 20–120 Hz range is immediately audible as an artifact even at levels too subtle to register on a gain reduction meter. Heavy low-band compression (more than 6 dB reduction, ratio above 4:1, release under 100ms) creates a sub range that feels artificially constrained rather than naturally controlled. The diagnostic is simple: if removing the low-band compression makes the track feel like it breathes again, you've over-compressed the foundation of the mix.
Multiband compression narrows stereo imaging at crossover frequencies, particularly with minimum-phase designs under heavy compression. Engineers who don't check stereo correlation before and after multiband insertion routinely deliver masters with imaging collapse in the upper mid range — frequencies between 2–5 kHz where stereo width is most perceptually significant. Use a stereo width meter and correlation meter on the multiband output, compare the width of the compressed signal to the uncompressed source at matched levels, and treat any width loss as a red flag requiring crossover or ratio adjustment.
Full bypass A/B comparison with multiband compression creates two problems: the gain difference between compressed and bypassed states makes the bypassed version sound thinner regardless of which sounds better, and bypassing all bands simultaneously prevents you from identifying which band is doing valuable work versus which is creating artifacts. The correct diagnostic is per-band bypass at matched output levels — engage each band individually, listen to what it changes, and evaluate whether the change is solving the identified problem or creating a new one.
The most common multiband compression mistakes all share a root cause — applying compression before completing a frequency-specific diagnosis of exactly which bands need independent dynamic control and why.
Red Flags and Green Flags
Red Flags
- 🔴 Over-compression in the low-mid band (200–500 Hz) hollowing out the body of vocals and instruments, creating a thin, nasal mix that loses power on small speakers
- 🔴 Mismatched release times causing pumping artifacts at band boundaries — especially between the low and low-mid crossover — audible as rhythmic wobbling that isn't groove-intentional
- 🔴 Using multiband compression to fix fundamental mix problems like clashing elements or poor gain staging instead of addressing those issues upstream — it will mask symptoms while making the root cause worse
Green Flags
- 🟢 Consistent low-end energy across the full duration of a track — the verse bass sits at the same perceived weight as the chorus without manual automation
- 🟢 High-frequency air and presence are preserved even as the low end is heavily tamed — the cymbals, vocals, and top-end detail remain open and unaffected by low-band compression activity
- 🟢 A mix that translates across systems (earbuds, club system, car stereo) with dramatically more consistency than it did pre-multiband, indicating balanced dynamic control across the frequency spectrum
Red flags in multiband compression consistently point to one of three root causes: crossover positions that don't match the source's frequency topology, ratio or threshold settings that are compensating for a problem that should be fixed upstream, or a decision to apply multiband where single-band compression or surgical EQ would have solved the issue more transparently. When you see pumping in a specific frequency range, stereo narrowing on loud sections, or a master that sounds tonally different at high volume versus low volume, the multiband configuration is the first place to investigate — specifically whether the band triggering the most gain reduction is responding to a real dynamic problem or a level problem that the mix never resolved.
Your Progression with Multiband Compression
Multiband compression rewards a staged learning approach because each level of proficiency unlocks a different class of problems the tool can solve. Trying to learn mastering-grade M/S multiband processing before understanding why a single low-band compressor on a bass guitar sounds different from a full-band compressor on the same bass is skipping the foundational knowledge that makes advanced technique legible. The progression below maps to real skills at each stage, not just increasing complexity.
Start by inserting a multiband compressor on a bass guitar or synth bass DI track. Identify the two most problematic frequency ranges — typically a boomy sub region below 80 Hz and a nasal upper-mid buildup around 300–500 Hz — and apply gentle compression (2:1, slow attack, moderate release) to only those bands, leaving all other bands inactive. Bypass each active band individually and listen for the difference. This teaches the core skill of multiband: hearing frequency-specific gain reduction in isolation rather than as part of an overall processed sound. Do not move to mix bus applications until per-band bypass produces an immediately audible and clearly beneficial result on individual sources.
Move to drum bus and mix bus applications with a three-band or four-band configuration. Practice setting crossovers by ear rather than by preset — play the track, watch the gain reduction meters, and move crossover points until each band is responding exclusively to its intended content. Introduce M/S awareness by monitoring stereo width before and after the multiband insert and adjusting settings if width loss is detectable. Study the interaction between attack time and crossover frequency by deliberately setting attack too fast and too slow in the low band, listening to what happens to kick transient character in each case. Begin comparing linear-phase and minimum-phase crossover modes on the same material to develop an ear for the tradeoffs.
Apply M/S multiband compression on final mix bus and mastering chains, compressing mid and side channels independently per band. Use dynamic EQ hybrid tools like FabFilter Pro-MB to target resonant problems at specific frequencies within a band rather than applying compression to the entire band's content. Develop a personal diagnostic workflow: spectrum analysis first, crossover placement second, gain reduction targets third, and per-band bypass verification last. Integrate LUFS metering into the multiband workflow to ensure that spectral balance improvements don't inadvertently shift integrated loudness in ways that affect streaming normalization behavior. At this level, multiband compression should be invisible — the listener hears a better mix, never the processor.
Progress through multiband compression by mastering per-band bypass diagnostics on individual sources before attempting mix bus applications, and reserve M/S multiband mastering for sessions where you can verify stereo integrity with correlation meters throughout the process.
Frequently Asked Questions
Regular (full-band) compression applies gain reduction uniformly across the entire frequency spectrum whenever the signal exceeds a threshold — a loud kick drum will trigger compression that also reduces the level of the hi-hats and vocals simultaneously. Multiband compression splits the signal into frequency bands first, so a loud kick will only trigger compression in the low-frequency band, leaving the mids and highs completely unaffected. This frequency-selective control is the defining advantage of multiband over single-band compression.
Multiband compression is most valuable on complex, broadband signals — mix buses, mastering chains, drum buses, and individual sources with problematic frequency-specific dynamics like bass guitar or full-mix stems. Applying it to every individual track is overkill and introduces unnecessary phase complications from the crossover filters. For most individual tracks, a well-set single-band compressor plus targeted EQ achieves cleaner results with less processing overhead.
Three to four bands is the professional standard for mastering — a sub/low band (up to 120–150 Hz), a low-mid/mid band (150 Hz–3–5 kHz), and a high band (above 3–5 kHz), with an optional split of the low-mids from the upper-mids at around 500–800 Hz. More bands introduce more crossover points, more phase complications, and more places where over-compression can occur. The Waves C4 and Fabfilter Pro-MB are both four-band tools used on countless commercial masters for exactly this reason.
Standard starting crossover points are: 120 Hz (sub/low split), 500 Hz (low/mid split), and 3.5–5 kHz (mid/high split). These align with natural frequency regions where problematic dynamics cluster — boomy sub content below 120 Hz, muddy buildup in the 200–500 Hz zone, and sibilance and harshness above 3.5 kHz. Always adjust crossovers based on the actual spectral content of your source rather than defaulting to preset values.
Thinness is almost always caused by over-compressing the low-mid band (200–500 Hz), which contains the fundamental body and warmth of most musical elements including bass, kick, vocals, and guitars. Even 6 dB of gain reduction in this range strips the fullness that gives a mix weight and density. Set your threshold so the low-mid band only compresses during the loudest moments, targeting 2–3 dB of gain reduction at peaks, and use makeup gain carefully to restore any level lost.
Both tools apply frequency-selective dynamic processing, but they differ in design philosophy. A dynamic EQ boosts or cuts a specific frequency band based on the signal level — it behaves like an EQ that responds to dynamics. A multiband compressor acts like a separate compressor on each band with ratio, attack, release, knee, and makeup gain controls that offer more precise dynamic shaping. In practice, dynamic EQ tends to sound more transparent and musical for subtle correction, while multiband compression offers more aggressive and metered control useful for mastering and heavy mix bus work.
Pumping in multiband compression is caused by fast release times relative to the tempo of the music — the compressor is releasing between notes or beats, creating an audible breath. Set release times so the gain reduction recovers in sync with musical phrases rather than individual transients: for a 120 BPM track, releases in the 300–600 ms range often feel natural in the mid and high bands. In the low band, releases below 80–100 ms are the usual culprit — try slowing the low-band release first, then check for band-boundary interactions at the crossover.
Yes — a single high-frequency band set to 5–10 kHz and above, with a moderate ratio (3:1–4:1), fast attack, and medium release, effectively functions as a broadband de-esser. This approach works well when sibilance is a systemic mix-wide problem. However, for individual vocal tracks, a dedicated de-esser or dynamic EQ targeting 5–8 kHz precisely will be more surgical and transparent than a multiband high-band compressor, which will also reduce legitimate high-frequency energy like air, presence, and cymbal shimmer alongside sibilance.