How to compress vocals (control, not crush)

Compression is what makes a vocal sit steady in a mix — every word audible, no line jumping out or disappearing. It's also the effect most people overdo, squashing the life out of a performance chasing "consistent." Here's how to compress a vocal for control instead of crush, and why two gentle passes usually beat one hard one.

Key Takeaways

• Compression evens out a performance so quiet words are heard and loud ones don't leap. Aim for control, not crush.
• A few dB of gain reduction on the peaks is plenty for most vocals.
• Two lighter compressors in series often sound more natural than one working hard.
• De-ess before you compress, so the compressor isn't lifting the esses.

What compression is actually doing

A vocal performance has a wide dynamic range: a whispered line and a belted one can be 20 dB apart. Left alone, the quiet parts vanish under the track and the loud parts poke out. A compressor turns the loudest parts down (and, with makeup gain, brings the whole vocal up), so the range narrows and every word sits in roughly the same place. That's the job. Everything else is taste.

A starting point for settings

Settings depend on the voice and the compressor, but here's where to begin:

• Threshold: lower it until you see a few dB of gain reduction on the loudest words, no more to start.
• Ratio: 2:1 to 4:1 for gentle, musical control. Higher ratios are for effect, not general leveling.
• Attack: medium. Too fast and you dull the consonants that give the vocal its attack; too slow and peaks slip through. Let a little transient through.
• Release: medium to fast, set so the compressor recovers between words without pumping.
• Makeup gain: bring the compressed vocal back to match the bypassed level so you're judging the compression, not the loudness.

Watch the gain reduction meter more than the numbers. A few dB on the peaks, moving musically, is the target.

Two gentle passes beat one hard one

A single compressor doing 8 dB of reduction has to work hard, and hard work shows up as pumping and lifelessness. Two compressors each doing 3 to 4 dB share the load and stay transparent: often a slower one first to catch the broad dynamics, then a faster one to tame remaining peaks. It's the same total control with half the strain on either.

De-ess before you compress

Here's a chain-order point that matters: de-ess before the compressor, not after. A compressor with a medium-fast attack lifts quiet detail, and that includes the front of every ess. Compress first and you've amplified the sibilance, so you then have to de-ess harder and dull more of the vocal.

De-ess first with a phoneme-aware tool like Sibilance, which only touches the ess, and the compressor works on a signal that's already smooth up top. It's why de-essing sits early in the vocal chain, and part of the larger picture in how to mix vocals.

Common compression mistakes

• Too much reduction, until the vocal is flat and lifeless instead of controlled.
• Attack too fast, dulling the consonants and the vocal's natural attack.
• Judging by loudness instead of matching levels, so "more compression" just sounds "louder and therefore better."
• One compressor doing all the work instead of sharing it across two gentle stages.

The short version

Compress for control, not crush: a few dB on the peaks, a musical attack and release, and two light passes rather than one heavy one. De-ess before the compressor so it's not lifting the esses, and match your levels so you're judging the compression honestly. A steady, natural vocal is the goal — and it starts with clean, de-essed input from Sibilance, free tier included.

Sibilance is the single-authority de-esser this comes from — a real free tier, and a 7-day Pro trial with no card.

Related reading