TL;DR
De-essing is compression on a narrow frequency band. Target 5 to 9kHz for most vocals. Threshold should only catch the harshest S sounds, not every sibilant. Use 3 to 6dB of reduction. Split-band mode beats broadband every time. And fix the mic/performance first if you can.
A great vocal with harsh sibilance sounds painful on headphones, amateur on earbuds, and unusable on small speakers. De-essing is the fix. Done right, the listener never notices it happened. Done wrong, the vocal loses life and every "s" sounds like a whispered "th."
This guide walks through what de-essing actually is, the settings that matter, the best plugins to use, and the five mistakes that make beginner de-essing worse than no de-essing at all.
What Sibilance Is and Why It's Brutal on Vocals
Sibilance is the high-frequency hiss produced by the letters S, Z, T, Sh, Ch, and F. It typically lives between 4kHz and 10kHz depending on the voice and mic.
Why it matters:
- Sibilant frequencies are the loudest part of the harmonic spectrum in most vocals
- Compression, EQ boosts, and reverb all emphasize sibilance further
- Bright mixes and bright plugins push it past acceptable
- On earbuds or cheap speakers, harsh S sounds become genuinely painful
The job of a de-esser is to briefly duck those specific frequencies only while a sibilant is happening. Not to cut them permanently (that's EQ's job). Not to squash everything (that's compression's job). Just to catch the spikes.
The Two Types of De-Essers
Broadband de-esser: Detects sibilance at a specific frequency band, then ducks the entire signal's volume. Easier to set up but pumps the vocal level on hard sibilants, which can be noticeable.
Split-band (or "dynamic") de-esser: Only ducks the sibilant frequency range, leaving the rest of the vocal untouched. Transparent, modern approach. Use this one 95% of the time.
Almost every modern de-esser plugin (FabFilter Pro-DS, Waves Sibilance, Oeksound Soothe2, Logic's built-in de-esser) defaults to split-band. If you have the option, keep it there.
The 4 Settings That Actually Matter
1. Target Frequency
This tells the de-esser where to look for sibilance. Different voices have sibilance in different bands.
| Voice Type | Start At | Adjust Range |
|---|---|---|
| Female / Soprano | 7 kHz | 6 to 9 kHz |
| Female / Alto | 6.5 kHz | 5.5 to 8 kHz |
| Male / Tenor | 6 kHz | 5 to 7.5 kHz |
| Male / Baritone, Bass | 5 kHz | 4 to 6.5 kHz |
Pro move: use the plugin's "listen" or "solo" mode if it has one. You'll only hear what the de-esser is catching. Sweep the frequency until the Ss jump out. That's your target.
2. Threshold
This is where the de-esser starts reducing gain. Set it so only the harshest sibilants trigger it, not every single "s" in the performance.
Too high: The de-esser barely triggers. Harsh sibilance slips through untouched.
Too low: Every "s" gets squashed. Vocal sounds lispy, like the singer has a head cold.
Right: The gain reduction meter flashes 3 to 6dB on the loudest Ss. Quieter Ss barely move the meter.
3. Ratio
How aggressively the de-esser reduces the triggered signal. Most sibilance problems are solved with 3:1 to 6:1.
- 2:1 to 3:1: Transparent, natural, leaves some sibilant character
- 4:1 to 6:1: The working range for modern pop
- 8:1 and up: Crushing. Use only when the source is genuinely harsh and you can't re-record
4. Range (on dynamic EQ de-essers)
On FabFilter Pro-DS and similar plugins, "Range" limits the maximum amount of reduction. Set this to 6 to 9dB so the de-esser can't over-reduce on a particularly harsh moment.
Where to Put the De-Esser in Your Chain
Signal flow matters. Put the de-esser in the wrong slot and you either miss sibilance or create it yourself.
Recommended vocal chain order:
- High-pass filter (remove rumble)
- Subtractive EQ (fix problem frequencies)
- Compressor (even out dynamics)
- De-esser (catch sibilance)
- Additive / tonal EQ (color and brightness)
- Saturation (optional)
- Second light compressor (glue)
The critical rule: de-ess before any additive high-frequency EQ boost. If you boost 10kHz first, the de-esser has to work twice as hard. If you de-ess first, your EQ boost doesn't push the sibilance back into harsh territory.
Some engineers put a second de-esser at the end of the chain for stubborn cases. That works, just don't use it as a fix for overly bright settings elsewhere in the chain.
