After Effects Beta 27.0x41 – Audio De-esser

Starting with After Effects Beta 27.0x41, a new Audio De-esser effect is now available. This compliments the previous audio effect updates released in 26.0 (e.g. compressor, noise gate, distortion).

What is a de-esser?

If you've ever recorded a voiceover or narration and noticed that the "s", "sh", "z", and "t" sounds jump out sharp and hissy — almost whistling — that's sibilance. It's completely normal, especially with certain microphones, close-mic'd voices, or bright EQ. But it can be fatiguing to listen to and makes an otherwise clean recording feel harsh.

A de-esser fixes exactly this. It's a smart, automatic volume control that only kicks in on those harsh high-frequency sounds and gently turns the the audio down automatically. You get a smoother, more polished vocal without dulling the whole track.

The new De-esser effect in After Effects lives under Effect > Audio > De-esser. Drop it on any audio layer and you're ready to go. 

Why use it?

• Vocals and narration — the most common use. Removes the "essy" harshness from spoken word.
• Podcasts and interviews — keeps long-form speech comfortable to listen to.
• Sung vocals — tames sibilance that gets exaggerated by compression or bright mixes.
• Anywhere a bright, hissy top end is distracting — even some dialogue or sound effects.

The goal is subtlety: a good de-esser is one you don't notice. You're aiming to take the edge off, not remove the "s" sounds entirely.

Getting Started

1. Add the effect to your audio layer (Effect > Audio > De-esser).
2. Turn on Sibilance Only and play the audio.
3. Adjust Frequency and Bandwidth until the "s" sounds are the loudest thing you hear.
4. Turn Sibilance Only back off.
5. Lower the Threshold until the harshness is gone — but stop before the voice starts to sound dull or lispy.
6. Tweak Release if you hear any pumping.

The Parameters
 

Threshold

This sets how sensitive the de-esser is. Lower (more negative) values make it more aggressive — it acts on more of the sibilance. Higher values (closer to 0) make it gentler. It only reacts to the loudest, harshest "s" sounds. Start in the middle and adjust by ear until the harshness is gone but the voice still sounds natural.

Frequency

This is where the effect goes looking for sibilance. Most "s" and "sh" sounds live in the 5,000–8,000 Hz range, and the default of 7,000 Hz is a great starting point. If the voice is deeper (many male voices), try nudging it lower (4,000–5,000 Hz). For brighter or higher voices, push it up (8,000–10,000 Hz).

Bandwidth

This controls how wide a slice of frequencies the effect targets around your chosen Frequency. A wider bandwidth catches more of the sibilant sound but can start affecting nearby parts of the voice. A narrower bandwidth is more surgical and less noticeable, but might miss some of the harshness. The default of 3,000 Hz works well for most voices.

Attack

How fast the effect clamps down once it hears a harsh sound. The default (1 ms) is nearly instant, which is usually what you want so it catches the sibilant as it happens. Slightly slower values let the very start of a word through before acting, which can sound more natural on some voices.

Release

How quickly the effect backs off after the sibilant passes. Too fast and you may hear a subtle "pumping" as it grabs and releases; a bit slower (the default is 50 ms) smooths that out. Longer values (100–200 ms) may give more a transparent, “invisible” result.

Sibilance Only

This is a setup helper, not something you leave on for the final mix. When enabled, you hear only the frequencies the effect is currently listening to. Turn it on, play your audio, and adjust Frequency and Bandwidth until the "s" sounds are loud and clear in your ears — that means you've zeroed in on the sibilance. Then turn it back off and let the de-esser do its job.

Suggested starting points

These are good places to begin but always trust your ears and adjust from here.

General voiceover / narration

• Threshold: around -20 dB
• Frequency: 7,000 Hz
• Bandwidth: 3,000 Hz
• Attack: 1 ms
• Release: 50–100 ms

Deeper male voice

• Threshold: around -18 dB
• Frequency: 4,000–5,000 Hz
• Bandwidth: 2,500 Hz
• Attack: 2 ms
• Release: 100 ms

Bright / higher voice

• Threshold: around -20 dB
• Frequency: 8,000 Hz
• Bandwidth: 3,000 Hz
• Attack: 1 ms
• Release: 75 ms

Just take the edge off (very subtle)

• Threshold: around -10 dB
• Frequency: 7,000 Hz
• Bandwidth: 2,000 Hz
• Attack: 1 ms
• Release: 100 ms

A Few Tips

• Less is more. If you can clearly hear the de-esser working, you've probably gone too far. Back off the Threshold.
• If words start sounding like a lisp, you're removing too much — raise the Threshold or narrow the Bandwidth.
• Every voice is different. The defaults are a starting point, not a rule. The Sibilance Only mode is the fastest way to dial it in for a specific recording.