Multiband Dynamics: change one region, protect the rest

Is the problem level-dependent?

If a part is always too bass-heavy, an EQ cut or a track-level change may be the simpler solution. Multiband Dynamics earns its complexity when a frequency region needs different treatment at different levels. It splits audio into Low, Mid and High bands, applies each band’s dynamics, then recombines them. Crossovers are overlapping filter transitions, not fences that perfectly isolate individual instruments.

Our original synthetic drums alternate low, pitched hits with noise-rich hits. The first lesson compresses only the Low band above its threshold. Low-Mid is 160 Hz and Mid-High is 2.5 kHz. In the Low row, Above T is −32 dB, Above R is 4, Att ms is 12 and Rel ms is 160. All Below R values, plus the other bands’ Above R values, stay at 1. There is no added band output gain.

Read one row before touching the others

Listen for less low-frequency body during the stronger hits while the noise-rich attack retains more independence than it would under full-band compression. Change the Low band’s Above R from 4 to 1. The above-threshold dynamics are now neutral; restore 4 to compare. These ratios are the device’s numeric R values, so enter 4 in the cell rather than typing a ratio string.

Next raise Low Above T from −32 to −20 dB. Less of the low-band envelope should need reduction. This is usually a better retreat than boosting Low Out dB to recover everything you just removed. If low notes still sound thin at ordinary levels, bring Above R closer to 1 or undo the band processing. The goal is a smaller difference between excessive and ordinary bass, not permanent bass removal.

The crossover is part of the decision

With the initial dynamics restored, move Low-Mid from 160 to 300 Hz. More upper-bass and low-mid material now participates in that band’s detector and gain movement. Listen for the hit body getting less solid, not just less boomy. Return to 160 Hz if the correction spreads too far. Moving it down to 90 Hz narrows the low-band task, but may leave the original problem above it.

Keep Low-Mid below Mid-High. The DSP constrains crossover ordering, so crossed settings are not a meaningful way to create an inverted band. Also remember that these crossovers change phase: neutral ratios do not mean the processed signal is sample-identical to bypass. Leave Dry/Wet at 100% during this experiment; blending a phase-shifted crossover path with dry audio can change tone independently of the compression.

Timing, Amount, and the settings left alone

Lengthen Low Rel ms from 160 to 400 to hear the bass stay reduced farther into the gap. Shorten it again if the next hit lacks weight. The global Time knob scales band attack and release together: 200% doubles both, so it is not equivalent to changing release alone. Keep Time at 100% when investigating a single timing control.

Lower Amount from 100% to 50% for a less pronounced dynamic correction. Amount scales input-gain and dynamic-gain contributions; it is not the same as mixing half the original waveform back in. With this example’s zero output and master gains, it is a useful final restraint. It should not excuse a poorly chosen crossover or an unnecessarily deep threshold.

Below-threshold processing is deliberately neutral here. Leave Below R at 1 until you have a specific reason to reshape quiet material; changing both regions at once hides cause and effect. Soft Knee smooths the transition, and RMS versus Peak changes detection, so compare those only after the basic correction works. This device has no separate sidechain EQ: external detection uses matching crossover bands. If several bands need constant rescue, revisit the mix instead of building three compensating problems.