Multiband compression: what it does and when to use it
Last updated 7 min readTags:MixingCompressionDynamics
Learn how multiband compression controls changing frequency ranges, when to choose it over EQ, and how to set crossovers, attack and release.
Multiband compression controls the dynamics of separate frequency ranges. You can reduce an uneven low end or tame bright cymbal hits without applying the same gain reduction to the rest of the signal.
It is useful when the tonal balance changes during a performance. If a recording is too bass-heavy throughout, ordinary EQ may be sufficient. If the bass becomes excessive only on certain notes, multiband compression can reduce those moments and leave the quieter notes largely alone.
How multiband compression works
A conventional multiband compressor uses crossover filters to divide the audio into bands, then compresses and recombines them. For example, three bands might cover below 150 Hz, 150 Hz to 3 kHz, and above 3 kHz. The boundaries depend on the recording.
Each band has its own threshold, ratio, attack and release. These work much as they do in full-band compression, but control separate frequency ranges.
The bands separate frequencies, not instruments. On a full mix, a low band can contain kick, bass and the lower notes of other instruments. All audio within that band receives its gain changes.
When to choose multiband compression
Choose the processor according to what needs to change:
| What you hear | What to try first |
|---|---|
| A recording is consistently too bright or bass-heavy | Static EQ |
| The whole performance has uneven levels | Full-band compression or volume automation |
| A broad frequency range becomes excessive only at certain moments | Multiband compression |
| One narrow resonance jumps out on particular notes | Dynamic EQ |
| Vocal “s” sounds are too strong | A de-esser, or a suitable high band |
| Kick and bass overlap, but the bass's upper harmonics sound fine | Compression on the bass's low band, triggered by the kick |
Dynamic EQ offers bell and shelf shapes for precise frequency adjustments. Multiband compression suits broader ranges that need separate dynamics settings. Their uses overlap; the number of bands alone does not determine which to choose.
If a kick makes a full-band mix compressor turn everything down, its detector may be too sensitive to the low end. A sidechain high-pass filter, if available, can reduce that sensitivity. This changes what triggers the compressor; it does not filter bass out of the audible signal.
How to set a multiband compressor
1. Choose the track or bus
Find the source of the problem. If one bass track has uneven low notes, process that track. Compression on the master would also affect the kick and other low-frequency content.
Loop a passage with both troublesome moments and material that already sounds right.
2. Set the frequency range
Use the band's solo function briefly to locate the unwanted sound and set its boundaries. Then turn solo off and judge the compression in context.
For excessive lows or highs, one crossover may suffice. A middle range needs two boundaries. Some plug-ins offer freely placed regions; others have fixed band layouts. Leave unneeded bands uncompressed, with their audio still audible.
Crossover slopes overlap around their boundaries. They do not isolate an instrument perfectly, so listen for useful parts of the sound that might also be reduced.
3. Set the ratio and threshold
For subtle correction, try 2:1, then lower the threshold until the band responds mainly to the excessive moments. Start with about 1–3 dB of gain reduction on those moments and adjust by ear. Stronger treatment may suit some recordings.
Use the range control, if available, to cap reduction. If the band stays heavily compressed, try a higher threshold or consider EQ.
4. Adjust attack
A shorter attack catches more of a sharp pick, consonant or cymbal hit. A longer attack lets more of the initial transient through before reducing the sustained sound.
Listen for lost impact. If the kick becomes weak or a guitar loses its articulation, slow the attack or reduce the compression. Very fast settings can also introduce distortion, particularly at low frequencies.
5. Adjust release
Release controls how quickly the band recovers from gain reduction.
If the level flutters or sounds distorted, try a longer release. If the band remains reduced into the next note or phrase, try a shorter one. A release that suits brief sibilance will not necessarily suit sustained bass notes.
Some plug-ins use adaptive timing or percentage controls rather than milliseconds. Judge the audible response and gain-reduction meter rather than copying a value from another compressor.
6. Compare at matched loudness
Match the overall loudness of the processed and bypassed versions. Check whether the uneven range is better controlled without losing the impact and tone you want.
Band make-up gain is optional and changes the tonal balance. If the result is dull, first check whether the high band is too broad or compressed too often.
Example: uneven low notes on a bass track
Suppose several low notes become boomy, while the rest of the bass part sounds balanced. As a starting experiment:
- Put the compressor on the bass track and set a low band below roughly 150 Hz.
- Leave the upper range uncompressed.
- Try 2:1, an attack of 25 ms and a release of 150 ms, if the plug-in uses milliseconds.
- Lower the threshold until the boomy notes produce around 2–3 dB of reduction and the other notes receive less.
Adjust the crossover to cover the excess without reducing useful midrange. Shorten the release if reduction carries into the next note. For one narrow ringing frequency, try dynamic EQ.
To make the low band dip when the kick hits instead, route the kick to the compressor's external sidechain and select it as that band's trigger. The bass is the processed audio; the kick supplies the control signal. Routing options vary by plug-in.
Can you use it on the whole mix?
Yes. It can stabilise tonal balance or alter a mix's dynamics before subsequent limiting. Creative uses may involve several bands and stronger settings than the corrective examples above.
The main limitation is shared frequency content. A high band intended to control cymbals may also reduce vocal consonants and guitar detail. If you still have the individual tracks, compare the result with processing the offending source directly.
There is no rule that four bands is too many. Each active region should improve the result at matched loudness.
Crossover filters and sound quality
Crossovers can affect phase and transient shape, depending on their design. Traditional minimum-phase filters introduce phase shifts. Linear-phase modes avoid relative phase shifts on recombination but add latency and can produce pre-ringing, an artefact heard before a transient. Some plug-ins use other approaches; FabFilter Pro-MB's Dynamic Phase mode, for example, avoids static crossover phase changes when no gain processing is applied.
A phase shift does not automatically mean audible damage. Compare the available modes on your material, particularly around kick and snare attacks. The linear-phase EQ guide explains the related trade-offs.
Common problems
| What you hear | What to check |
|---|---|
| Bass loses weight or punch | Reduce low-band gain reduction; try a slower attack |
| Vocals or cymbals become dull | Raise the high-band threshold, reduce the ratio or narrow the affected range |
| The level audibly swells and dips | Adjust release and check whether too much reduction is being applied |
| The stereo image shifts | Check whether left and right channels are compressed independently; try more stereo linking |
| Attacks sound smeared or have a faint sound before them | Compare crossover modes and check for linear-phase pre-ringing |
| The improvement disappears after level matching | Reconsider the settings; extra loudness may have influenced the comparison |