Upward compression and expansion explained
Last updated 7 min readTags:MixingCompressionDynamics
Upward compression raises quiet sounds; expansion increases level differences. Learn when to use each and how to set them without excess noise or lost tails.
Upward compression raises audio below a threshold. It can make quiet notes, drum sustain and reverb tails more audible without directly reducing louder peaks. It also raises any noise present in those quiet moments.
Expansion increases the difference between loud and quiet. A downward expander turns quiet audio down, often to reduce noise or microphone bleed between notes. An upward expander raises audio above a threshold, which can give stronger hits more emphasis.
The choice depends on what needs to change: quiet detail that gets lost, unwanted sound between notes, or hits that need more impact.
The four types of dynamics processing
Compression reduces level differences; expansion increases them. The words upward and downward describe whether the processor adds or removes gain.
| Process | Acts on audio | Level change | Typical use |
|---|---|---|---|
| Downward compression | Above the threshold | Turns it down | Controlling loud vocal phrases or bass notes |
| Upward compression | Below the threshold | Turns it up | Raising quiet notes, sustain or room sound |
| Downward expansion | Below the threshold | Turns it down | Reducing noise or bleed between notes |
| Upward expansion | Above the threshold | Turns it up | Emphasising stronger drum hits |
These are the basic gain behaviours, before output gain and timing settings are considered. A soft knee makes the transition gradual around the threshold. A multiband processor can apply different behaviours to separate frequency bands.
Ableton's Multiband Dynamics manual illustrates all four types.
When upward compression is useful
Upward compression suits a recording whose louder parts already sit well in the mix, but whose quieter details are hard to hear. On acoustic guitar, it might bring out soft picking between stronger chords. On drums, it can raise the decay and room sound after each hit.
For example, with a threshold of −24 dBFS, audio detected at −30 dBFS is below the threshold and can receive a boost. Audio at −12 dBFS is above it and receives no upward boost in the basic static curve. The amount of boost depends on the ratio and any gain limit; attack, release and knee affect the response in practice.
It cannot distinguish quiet music from quiet noise. Hiss, breaths, finger squeaks and microphone bleed may become more prominent too. Within a passage receiving the same gain, the wanted sound and its background noise rise together.
If only one word or note needs attention, use clip gain or automation. That gives you control over the specific moment without raising every quiet passage. The dynamic range control guide covers these options.
How to set an upward compressor
- Loop a passage with both quiet and loud material. Listen in the mix so you can judge how much extra detail is needed.
- Set the threshold above the quiet material you want to raise. Keep it below louder material you want to leave alone. Watch the gain display to confirm which passages receive a boost.
- Use gentle compression and limit the boost. If the plug-in has a range control, try a maximum lift of 2–4 dB as an initial audition. This is a listening test, not a target for every track.
- Adjust the timing. Listen for noise swelling between notes, exaggerated breaths or an unnatural rise in the decay. Reduce the amount or change the timing if these become distracting.
- Match the overall listening level and bypass. Keep the processing only if the quiet material is easier to hear and the performance still sounds natural.
Ratio displays and timing controls vary between processors. Check the manual rather than assuming that a setting behaves exactly as it does on a conventional compressor. FabFilter's Pro-C 3 documentation, for instance, identifies a dedicated Upward style.
Is parallel compression the same thing?
Parallel compression blends a compressed copy with the dry signal. It can produce a similar increase in quiet detail relative to peaks, but the result depends on the compressor settings and blend.
The dry path retains the original waveform; the combined output can still have higher peaks. Check the level after the blend. Neither approach guarantees that the final peak level will stay unchanged.
Downward expansion: reducing sound between notes
A downward expander reduces gain when the detected level falls below its threshold. On a snare microphone, it can lower hi-hat bleed between snare hits. On a vocal, it can reduce room noise in pauses.
It does not remove noise underneath louder wanted sound while the expander is open. If the noise and wanted notes are at similar levels, a threshold alone will not separate them reliably.
A gate applies a more abrupt or stronger reduction. An expander is often preferable when you want quieter gaps but still need some room sound and natural decay. The distinction depends on the ratio, range and timing, rather than the label alone.
The range control limits attenuation. A maximum reduction of 10 dB makes the gaps quieter without muting them. A much deeper range can make the track fall almost silent between notes. FabFilter Pro-G's manual explains how its ratio and range work together.
How to set a downward expander
- Loop the quietest wanted notes and the gaps around them. A threshold that works for strong hits may suppress softer ones.
- Place the threshold above the unwanted sound and below those wanted notes. If there is no useful separation, try editing or automation instead.
- Start with a gentle ratio and limited range. Try 6–12 dB of maximum reduction as an initial audition; complete silence is not always desirable.
- Check the start of each note. Shorten the opening time, or use lookahead if available, when the first part of a hit is lost.
- Check the decay. Adjust hold and release so notes finish naturally. Rapid opening and closing, often called chatter, may need a longer hold or release, or a different threshold.
Judge the result across the whole performance. Listen especially for missing ghost notes, clipped word endings and cymbal or reverb tails that stop abruptly.
Upward expansion: emphasising louder hits
Upward expansion adds gain above a threshold. It may suit a drum loop where stronger hits need more emphasis relative to the quieter material. It also increases peak levels, so leave headroom and check for clipping afterwards.
It is different from transient shaping. An upward expander responds to level relative to a threshold; a transient shaper responds to the sound's attack and sustain. A quiet hit may need more attack even though it never crosses an expander's threshold.
Expansion can improve the perceived impact of some compressed material, but it cannot reliably reconstruct the original transients after heavy limiting or clipping. If you have the session, reduce the earlier processing before trying to compensate for it.
Which process should you try?
| What you hear | First option to try | Listen for |
|---|---|---|
| Quiet details disappear, but louder parts sit well | Upward compression | Extra noise, breaths or exaggerated sustain |
| Bleed or room noise is intrusive between notes | Downward expansion | Missing soft notes or shortened tails |
| Loud phrases jump out of the mix | Downward compression | Dulled attacks or audible level movement |
| Strong drum hits need more emphasis | Upward expansion | Excessive peaks or too much contrast between hits |
| One word or note is too quiet | Clip gain or automation | A natural level change into and out of the edit |
Make the decision in the mix. A snare that sounds tidier in solo may lose useful room sound; a guitar with more audible detail may also draw attention to every finger movement. Keep the smallest adjustment that solves the audible problem.