Multiband Compression
A multiband compressor splits the signal into frequency bands — typically three to five — and applies independent compression to each band. The low end can be compressed heavily while the highs stay dynamic, or vice versa.
How It Works
Vocabulary
Crossover Network
A set of filters that splits a signal into separate frequency bands for independent processing. The same principle used inside speakers, graphic EQs, and multiband compressors.
Crossover points divide the spectrum into bands — a crossover network. A three-band compressor might split at 200 Hz and 3 kHz — everything below 200 Hz goes to the low band, 200 Hz to 3 kHz goes to the mid band, everything above 3 kHz goes to the high band. Each band has its own threshold, ratio, attack, and release.
After compression, the bands are recombined into a single output. The result: frequency-specific dynamics control that a broadband compressor can’t achieve.
Here’s an exercise that reveals what a multiband compressor really is: take any signal and split it into three bands using three copies with different EQ settings — one low-passed at 200 Hz, one band-passed between 200 Hz and 3 kHz, one high-passed at 3 kHz. Put a separate compressor on each. Blend them back together. Congratulations — you just built a multiband compressor from scratch. And if you remove the compressors, you’ve built a crossover network — the same thing inside a graphic EQ, a speaker system, and a multiband processor. It’s all the same principle: split by frequency, process independently, recombine.
Signal split at crossover points into three frequency bands, each with its own compressor, then recombined.
When to Use It
- Taming a boomy low end without affecting the vocal clarity. A broadband compressor triggered by a bass-heavy kick would also compress the midrange — ducking the vocal on every kick hit. A multiband compressor limits the low band only.
- Controlling harsh vocal sibilance without dulling the entire voice. Compress the high band to catch the “ess” sounds while leaving the fundamental vocal tone untouched. (A de-esser, covered below, does this more specifically.)
- Mastering. Multiband compression on the master bus lets you control the energy of each frequency range independently — tightening the bass without squashing the transients of the snare and cymbals.
The Risk
Multiband compression is easy to overcook. Each band is changing the level independently, which means the spectral balance of your signal changes depending on how loud it is. Heavy multiband compression can make a mix sound processed, unnatural, and “phasey” — the frequency balance shifts unnervingly between loud and quiet moments. Use it surgically, not broadly.
De-Essing
A de-esser is a specialized compressor designed to catch sibilance — the sharp “ess,” “tee,” and “shh” sounds in vocals that can become painful on certain microphones and in certain mixes.
There are two approaches:
Sidechain-based de-essing: The compressor listens to a specific frequency range (usually 4-10 kHz, where sibilance lives) but compresses the entire signal when sibilance is detected. Simple and effective, but it can dull the vocal momentarily on each “ess.”
Multiband de-essing: The compressor only reduces the gain in the sibilant frequency range, leaving the rest of the vocal untouched. More transparent, but requires more careful tuning to avoid affecting non-sibilant content in the same range.
Set a de-esser by finding the frequency where the sibilance is most prominent (solo the sidechain or sweep a narrow EQ boost to find it), then set the threshold so the de-esser catches the sibilant peaks without acting on normal vocal content. If the de-esser makes the singer sound like they have a lisp, it’s set too aggressively.
Notice something: these two de-essing approaches mirror the two approaches we’ve been learning. The sidechain-based de-esser is like a gate — it listens to one frequency range but acts on the whole signal. The multiband de-esser is like a multiband compressor — it only acts on the problematic frequency range. Same problem, two philosophies. Understanding when each approach is better is the difference between surgical mixing and blunt-force processing.
Two-panel de-esser comparison: sidechain (listens to HF, compresses full signal) vs multiband (only compresses sibilant band). Vocal waveform with ess peaks highlighted.
Dynamic EQ
A dynamic EQ combines an equalizer with a compressor — each EQ band only acts when the signal in that band exceeds a threshold. Below the threshold, the EQ band does nothing. Above it, the band applies its boost or cut.
This solves a fundamental problem with static EQ: a frequency range that’s problematic on loud notes might be fine on quiet ones. Static EQ cuts all the time. Dynamic EQ cuts only when needed.
Dynamic EQ vs. multiband compression: They’re similar, but dynamic EQ gives you more surgical control. A multiband compressor splits the signal into wide bands with crossover points. A dynamic EQ lets you target a specific narrow frequency range with surgical precision — a 200 Hz boost at Q of 4.0, activated only when the signal is loud enough.
Use dynamic EQ when the problem is narrow and intermittent. Use multiband compression when the problem is broad and consistent.
Dynamic EQ plugin showing a band that activates only when signal exceeds threshold — before/after the threshold being crossed.
Upward vs. Downward: The Other Dimension
So far, we’ve mostly talked about downward compression and expansion — loud things get turned down, quiet things get turned further down. But there’s another direction.
One question keeps all four modes straight: what gets left unaltered? Downward compression turns loud stuff down — quiet signals pass through untouched. Upward compression turns quiet stuff up — peaks pass through untouched. Downward expansion (gating) pushes quiet stuff further down — loud signals pass through untouched. Upward expansion pushes loud stuff further up — quiet signals pass through untouched.
The common explanation — “compression makes loud stuff quieter and quiet stuff louder” — is misleading. What actually happens: compression turns loud stuff down, and then you apply makeup gain to turn everything back up. The quiet stuff gets louder as a side effect of the makeup gain, not because the compressor touched it.
Upward compression makes quiet signals louder without affecting the loud parts. It brings up the details — the room ambience, the sustain of a guitar, the breath between vocal phrases — without touching the peaks. This is like parallel compression (Chapter 19) but done in a single process.
Upward expansion makes loud signals even louder — adding dynamic range by pushing the peaks further up. This is useful for restoring dynamics to over-compressed material.
These are powerful tools, but they’re less common in everyday mixing than their downward counterparts. Most compressors default to downward compression. When you encounter an “upward” mode or a “compressor” that brings up quiet signals, now you know what it’s doing. (The Upward and Downward Dynamics video breaks this down with visual examples.)
Sidechain Compression in Practice
Chapter 15 introduced sidechaining conceptually. The most common application: kick-duck-bass — using the kick drum to trigger compression on the bass track.
The setup: put a compressor on the bass track. Set the sidechain input to the kick drum track. Now, every time the kick hits, the compressor engages and pushes the bass down. Between kicks, the bass returns to its normal level.
Why this works: the kick and bass occupy similar frequencies. When they play simultaneously, they compete for the same space — the low end gets muddy and the kick loses definition. Sidechain compression creates a momentary pocket: the bass dips out of the way for each kick hit, then fills back in. The kick punches through clearly, the bass stays present, and the low end stays clean.
Settings for kick-duck-bass: moderate ratio (3:1 to 4:1), fast attack (catch the kick immediately), medium release (let the bass return between hits), and just enough gain reduction (3-6 dB) to create the pocket without making the ducking obvious.
Kick drum sidechain input to compressor on bass track. Waveform showing kick hits and bass ducking in response.
In electronic music, sidechain compression is often used more aggressively as a creative effect — the entire mix ducking on each kick hit creates the characteristic “pumping” feel of house and EDM.
What to Practice
- Set up sidechain compression on a bass track. Create a compressor on the bass, sidechain it to the kick. Adjust the threshold until you see 3-6 dB of gain reduction on each kick hit. A/B the compressed and uncompressed bass — listen for how the kick punches through more clearly with the sidechain active.
- De-ess a vocal. Find a vocal with harsh sibilance. Add a de-esser, find the sibilant frequency, and set the threshold. Reduce until the “ess” sounds are tamed but the vocal still sounds natural. Push it too far on purpose to hear what over-de-essing sounds like.
- Compare dynamic EQ to static EQ. Pick a track with an intermittent frequency problem — a guitar that gets boomy only on certain chords, or a vocal that gets harsh only on loud phrases. First, fix it with static EQ (a permanent cut). Then try dynamic EQ (a cut that only engages on the loud moments). Listen to which approach preserves more of the natural tone.
