Common De Esser Mistakes to Avoid

By Sophie Clarke 14 January 2025 3 min read

Filed under

Add Music Adviser UK to your preferred sources in Google

A de-esser is a focused tool, but several common setup errors can make it counterproductive or inaudible. The problems below appear regularly, are straightforward to identify and are each resolved by adjusting a single control.

Using a De-Esser When You Don't Need One

Not every vocal recording has a sibilance problem. Using the processor on a naturally warm or dark voice, or on a vocal recorded at a distance from the microphone, can remove high-frequency energy that is contributing positively to the vocal's presence. Before adding any processing, listen through the vocal critically and check whether harsh consonants are actually distracting — if not, leave the channel clean.

Targeting the Wrong Frequency

Sibilance peaks at different frequencies on different voices and microphones. Setting the detection frequency to 8 kHz and leaving it there may miss sibilance on voices that peak at 5 kHz or 12 kHz. Sweep the detection frequency slowly while the vocal plays and listen for the point where the processing catches the most sibilance without affecting voiced consonants and other articulation. Checking the frequency with a spectrum analyser before setting the control speeds up this process considerably.

Setting the Threshold Too Low

A threshold that is too low causes continuous gain reduction throughout the vocal, not just on sibilants. The result sounds lispy — the vocal loses natural brightness and clarity on every consonant, not only the harsh ones. Set the threshold so that the gain reduction meter moves only on genuine sibilant moments; in most cases, gain reduction should not be constant. The de-esser should be largely inactive during sung vowels and musical phrases between consonants.

Not Checking in Context

Processing calibrated in solo can appear to work correctly but behave differently in a full mix. Sibilance that sounds acceptable in isolation may become more or less prominent when other elements — cymbals, synthesisers, guitars — occupy similar frequency ranges. For setup guidance on the full process, the beginner's guide covers each step in sequence.

Frequently Asked Questions

What does over-processing with a de-esser sound like?

The vocal sounds lispy — the ability to produce clear s and sh sounds is reduced, the voice loses natural brightness, and the processing becomes audible as a side effect rather than a correction. When this happens, raise the threshold until gain reduction only occurs on the loudest sibilant peaks.

Why is the reduction not working even though it is active?

The threshold may be set too high for the level of sibilance in the recording, or the detection frequency may not be targeting the correct part of the spectrum. Check the gain reduction meter while the vocal plays — if it is not moving on sibilant consonants, lower the threshold or sweep the detection frequency.

Should I place it before or after compression?

Placing it before compression avoids the compressor reacting to sibilant peaks and raising the overall gain. Placing it after compression addresses sibilance that the compressor has emphasised by reducing the surrounding level. Both positions are valid; try both and choose the one that produces the most transparent result on the specific recording.

Is it better to automate volume than to use this kind of processing?

Volume automation or clip gain reduction on sibilant syllables is an alternative for recordings where only one or two consonants are problematic. It requires more editing time but can produce very clean results. For vocals with consistent sibilance throughout, dynamic processing is more practical.

Can it fix recordings that were too close to the microphone?

Proximity effect from close-miking boosts low frequencies rather than high ones, so it does not directly cause sibilance. However, on-axis miking at close range does increase high-frequency capture. Sibilance from this cause can be reduced with frequency and threshold adjustments, though re-recording at a greater distance would address the problem at the source.