Threshold and Ratio
These are the two fundamental controls. Everything else is refinement.
Vocabulary
Threshold
The level above which a dynamics processor begins acting. Below the threshold, the signal passes through unchanged. The threshold determines how much of the signal gets processed — set it high and only the loudest peaks are affected; set it low and the processor works constantly.
Threshold: The level above which compression begins. Signals below the threshold pass through unchanged. Signals above the threshold get turned down. Set the threshold high and only the loudest peaks get compressed. Set it low and the compressor is working constantly.
Vocabulary
Ratio
How much the compressor reduces the signal above the threshold. At 4:1, for every 4 dB the signal exceeds the threshold, only 1 dB comes through. Higher ratios mean more compression. Infinity:1 is a limiter — nothing gets above the line.
Ratio: How much the signal gets turned down. A ratio of 4:1 means that for every 4 dB the signal exceeds the threshold, only 1 dB comes through. A ratio of 2:1 is gentle — the signal is squeezed slightly. A ratio of 10:1 is aggressive — the signal is severely limited. A ratio of ∞:1 (infinity to one) is a limiter — nothing above the threshold gets through.
Example: threshold at -20 dBFS, ratio 4:1. A signal peak at -10 dBFS is 10 dB above the threshold. At 4:1, those 10 dB become 2.5 dB above the threshold. The peak that was at -10 dBFS is now at -17.5 dBFS. The compressor reduced the signal by 7.5 dB.
The amount of reduction is called gain reduction. Most compressors show this on a meter — watch it move while the compressor works. The meter tells you how much the compressor is doing at any moment.
One more parameter worth knowing: the knee controls how abruptly the compressor engages at the threshold. A hard knee means full ratio the instant you cross the line; a soft knee eases into compression gradually. We’ll explore this in the next chapter. (The Why Does a Compressor Have a Knee video demonstrates this visually.)
Compression transfer function graph showing threshold knee with lines at different ratios (2:1, 4:1, 10:1, ∞:1), worked example annotated.
Attack and Release
These control the speed of the compressor — not “how long,” but “how fast.”
Attack: How quickly the compressor reacts after the signal crosses the threshold. A fast attack clamps down on transients immediately — the initial “hit” of a drum, the pick attack of a guitar, the consonant at the start of a vocal phrase. A slow attack lets the transient through before the compressor engages, preserving punch and impact.
Release: How quickly the compressor lets go after the signal drops back below the threshold. A fast release means the compressor resets quickly, ready for the next peak. A slow release means the compressor continues to hold the signal down even after the loud moment passes, creating a smoother, more sustained character.
The interaction between attack and release defines the feel of the compression. Fast attack + fast release = aggressive, pumping. Slow attack + slow release = smooth, transparent. Fast attack + slow release = squashed, sustained. Slow attack + fast release = punchy, dynamic.
Drum hit waveform with two overlaid gain reduction curves: fast attack (catches transient) vs slow attack (lets transient through).
Makeup Gain
Compression turns loud things down. That means the overall signal is quieter after compression. Makeup gain adds level back after compression, bringing the signal up to its original perceived loudness.
This is where the deception comes in. If you add 6 dB of makeup gain and the signal sounds “better,” you need to check: is it actually better, or just louder? Bypass the compressor and compare at matched levels. If the compressed version is louder, turn the makeup gain down until they match. Then compare.
Three-panel before/after: original dynamic signal, compressed signal (quieter), compressed + makeup gain (restored level). Shows why level-matching matters.
Compression vs. Gain Automation
Not everything needs a compressor. A vocal that’s mostly even but dips for one phrase? Automate the fader up for that phrase. A guitar that’s consistent except for one loud chord? Clip-gain that chord down. These are surgical fixes for specific moments — compression is a continuous process that affects the entire signal.
The rule of thumb: if the problem is a few specific moments, use automation. If the problem is an overall lack of consistency, use compression. If both, do the automation first — fix the specific problems so the compressor doesn’t have to work as hard.
What to Practice
- Compress a drum loop. Start with a moderate ratio (4:1) and pull the threshold down until you see 3-6 dB of gain reduction on the peaks. Listen to how the character changes. Now adjust the attack: hear the difference between a fast attack (transients get squashed) and a slow attack (transients punch through).
- Level-match a compressed signal. Add a compressor to any track. Adjust until you see gain reduction. Add makeup gain to match the bypassed level. Now bypass and compare. Did the compressor actually improve the sound, or just change it? This is the most important exercise in this chapter.
- Automate instead of compressing. Take a vocal track with a few quiet moments and a few loud moments. Instead of compressing, ride the fader (or use clip gain) to even out the dynamics manually. Listen to how transparent this sounds compared to compression. Understand when each tool is appropriate.
