De-Essing Vocals: How to Fix Sibilance

Sibilance—those harsh, piercing "s" and "sh" sounds in vocals—is one of the most common problems in vocal mixing. Every recording has some. Some recordings have way too much.

A de-esser fixes it. But use it wrong and your singer sounds like they have a lisp.

Here's how to de-ess properly.

What Sibilance Actually Is

When a singer makes "s," "sh," "ch," or "t" sounds, air rushes through a narrow gap in their teeth and tongue. This creates high-frequency noise concentrated around 5kHz to 8kHz.

In a controlled live setting, you barely notice it. But microphones—especially condenser mics—pick it up harshly. Add some compression and EQ boost in the highs, and suddenly those sounds are stabbing your ears.

That's sibilance. It's not a recording mistake—it's just physics. Every vocal recording has it. The question is how much, and whether it needs treatment.

What a De-Esser Does

A de-esser is basically a compressor that only triggers on sibilant frequencies.

It listens for energy in the sibilance range (usually 5-8kHz). When it detects it, it briefly ducks those frequencies. When the sibilance passes, it returns to normal.

The result: the "s" sounds get tamed, but the rest of the vocal stays untouched.

Finding the Problem Frequency

Every voice is different. Male vocals typically have sibilance centered around 5-6kHz. Female vocals are often higher, around 7-8kHz.

To find it:

Loop a section with lots of sibilance
Use an EQ with a narrow band
Boost that band 6-10dB
Sweep through 4kHz to 10kHz
Listen for where the harshness peaks
That's your target frequency

Most de-essers let you solo the sidechain so you can hear exactly what the de-esser is targeting. Use this. It makes dialing in the frequency way easier.

De-Esser Settings

Frequency

Set this to the sibilance frequency you found. Usually 5-8kHz. Don't guess—find it.

Threshold

This determines when the de-esser kicks in. Set it so it only triggers on the actual sibilant sounds, not on everything.

Watch the gain reduction meter. It should only move when you hear an "s" or "sh" sound. If it's moving constantly, your threshold is too low.

Range/Reduction

How much the de-esser pulls down the sibilance when triggered. Start around 3-4dB. Go up to 6-8dB for harsh sibilance. Beyond 10dB starts sounding like a lisp.

Attack/Smoothing

Most de-essers have some kind of speed control. Faster catches the sibilance instantly but can sound unnatural. Slower is more transparent but might let some harshness through.

Start around 10ms and adjust.

Wide-Band vs. Split-Band

Wide-Band Mode

When triggered, the de-esser turns down the entire vocal signal—all frequencies. This sounds more natural but affects more than just the sibilance.

Split-Band Mode

When triggered, the de-esser only turns down the sibilant frequency range. The rest of the vocal stays at full level.

More transparent in theory, but can sound weird if you're cutting too much—the "s" sounds hollow while the rest of the word stays full.

My recommendation: Use wide-band for most situations. It sounds more natural. Only use split-band if wide-band is giving you problems.

Where to Put the De-Esser

There's debate about this. Here are the two main approaches:

Before Compression

The de-esser catches sibilance before the compressor sees it. This means the compressor won't react to (and potentially emphasize) the harsh sounds.

This is my usual approach.

After Compression

Sometimes compression brings out sibilance that wasn't obvious in the raw vocal. De-essing after catches anything the compressor made worse.

After EQ

If you're boosting high frequencies for "air," that boost will make sibilance worse. De-essing after your EQ catches the boosted sibilance.

The Two-Stage Approach

For really problematic sibilance, use two de-essers in series:

First de-esser: Gentle settings, catching the worst offenders (3-4dB reduction)

Second de-esser: Even gentler, cleaning up what's left (2-3dB reduction)

Two gentle passes sound more transparent than one aggressive pass. Each de-esser does less work, so neither sounds obvious.

Manual De-Essing

Some engineers skip de-esser plugins entirely and do it manually:

Go through the vocal
Find every sibilant sound
Select just that moment
Turn it down 3-6dB with clip gain

It's more work, but you have complete control. Every "s" gets exactly the treatment it needs—no more, no less.

For critical vocals where the de-esser isn't quite right, this is worth the effort.

Common Mistakes

Targeting the Wrong Frequency

If your de-esser is set to 5kHz but the sibilance is at 7kHz, it's not going to help. Find the actual frequency first.

Over-De-Essing

Too much reduction and your singer sounds like they have a lisp. Every "s" becomes a "th." Back off until the sibilance is controlled but the consonants still sound natural.

Using EQ Instead of a De-Esser

A static EQ cut at 6kHz affects the entire vocal, making everything in that range quieter—not just the sibilance. The vocal sounds dull. Use a de-esser so only the sibilant moments get treated.

Ignoring It Entirely

Some engineers leave sibilance alone, thinking it'll be fine. It won't. Sibilance that's borderline in mixing becomes unbearable after mastering adds brightness and loudness.

When You Need More Than De-Essing

Sometimes the sibilance is baked into a bad recording—cheap mic, wrong position, overly bright preamp.

In those cases, you might need:

Multiband compression targeting the sibilance range
Dynamic EQ for more surgical control
Multiple de-essers in series
Manual de-essing on the worst offenders
Re-recording (if possible and it's really bad)

The Quick Settings

For most vocals, start here:

Frequency: 6kHz (adjust to taste)
Threshold: Set so it only triggers on sibilance
Reduction: 4-6dB
Mode: Wide-band
Position: Before or after compression

Listen for harshness on the "s" sounds. Turn up the reduction until they're tamed. Back off if the singer starts sounding like they have a lisp.

That's it. De-essing isn't complicated—it just requires finding the right frequency and applying the right amount of reduction.