AUD · Audio tools

Multiband Compressor

A single compressor reacts to whatever is loudest, which on a full mix is usually the low end. Every bass note then turns down the entire track. Splitting the signal into bands first keeps that reaction where it belongs.

The three bands

Low split and high split set the two crossover points. Everything under the low split is the low band, everything over the high split is the high band, and what is left is the mid.

Each split is a low-pass filter and a subtraction rather than a pair of matched filters, so the bands reconstruct the input exactly when their gains are equal. There is no ripple at the crossovers to work around, and moving a crossover during playback ramps rather than steps.

Thresholds

Each band has its own threshold in dBFS, and each has its own detector linked across channels so the stereo image does not move when one side is louder.

Set them one at a time. Drop the low threshold until the bass stops dominating, leave the mid threshold high enough that vocals are only touched on peaks, and use the high threshold for cymbals or sibilance that push past everything else.

Ratio, attack, release

Ratio, attack, and release are shared by all three bands. Ratio is how much of the excess is held back: 3:1 means 12 dB over the threshold comes out 4 dB over.

Attack and release run from 1 to 100 ms and 20 to 1000 ms. Short attacks catch transients but flatten drums. Long releases keep the gain steadier and are more forgiving on a mix bus.

Makeup gain

Compression lowers the peaks, so the file gets quieter. Makeup gain adds up to 18 dB back at the output, after the bands are summed.

Add roughly as much as the loudest band is losing, then compare against the bypassed signal at the same loudness. Anything that only sounds better because it is louder is not better.

Example: a boomy podcast mix

Low split 140 Hz, high split 3 kHz. Low threshold -26 dB, mid -16 dB, high -20 dB. Ratio 3:1, attack 20 ms, release 250 ms, makeup 4 dB.

Room boom under 140 Hz is held back on every plosive, and the voice above it keeps its own dynamics.

Frequently Asked Questions

It splits the file into three frequency bands, compresses each one against its own threshold, and adds them back together, so a loud kick drum no longer turns down the vocal above it. It reads WAV, MP3, M4A, AAC, Ogg, Opus, FLAC, AIFF, and WebM audio, and exports in the format the source arrived in.

Use it when one part of the spectrum is causing all the gain reduction. Boomy bass, a harsh upper mid, or a loud sibilant top will pump the whole mix through a single-band compressor. If the level problem is spread evenly, a single-band compressor is the simpler choice.

180 Hz and 2.8 kHz split a mix into weight, body, and detail, which works for most material. Move the low split down to 80 to 120 Hz to isolate only the sub range, and pull the high split down to around 1.5 kHz to catch upper mid harshness.

No. Each band is one low-pass filter plus a subtraction, so the three bands always add back up to the input sample they came from. At a ratio of 1:1 with no makeup gain the output is the source file.

The attack or release is too short for the frequencies in that band. Low frequencies need longer times because one cycle at 60 Hz lasts about 17 ms, so try 30 ms attack and 250 ms release before changing the threshold.

Explore Our Tools

Browse all tools