Taming sibilance in vocals, step by step
Loud esses tire the listener and get worse after compression and streaming encoding. The right de-esser pulls them down without dulling the vocal.
Sibilants are the s, z and sh sounds: bursts of energy between 4 and 10 kHz. Compression, saturation and presence EQ make them grow, and lossy encoding on streaming platforms makes them harsher still. That is why the de-esser goes after compression and before the reverb.
The procedure
Find the frequency
Boost a narrow bell by 8 to 10 dB and sweep it between 4 and 10 kHz on a line full of esses. Where the s stings your ear, that is the frequency. Male vocals often sit between 4 and 7 kHz, female vocals between 6 and 9 kHz.
Set the de-esser to that frequency
Use the sidechain listen function, if there is one, to hear only what the de-esser detects. You should hear only the sibilants, not the vowels.
Choose the mode
Split-band mode lowers only the sibilance zone: it is the most transparent choice. Wideband mode turns the whole vocal down for an instant: fine for very harsh esses, but on dense vocals it can make the phrase stumble.
3 to 6 dB of reduction
Lower the threshold until the loudest esses drop by 3 to 6 dB. If the singer starts to sound like they have a lisp, you went too far.
Listen in the mix
A de-essed vocal always sounds a bit dull in solo. In the mix, with cymbals and synths around it, it should come across clear and smooth.
Fix the worst esses by hand
The two or three sibilants the de-esser cannot tame get lowered with clip gain, syllable by syllable. It is the most precise and least audible method.
De-ess the reverb too
Sibilants in the reverb hang in the air after the word. A de-esser on the reverb send, or a low-pass on the return, removes them.
Starting values
| Vocal | Frequency | Reduction |
|---|---|---|
| Male | 4 to 7 kHz | 3 to 5 dB |
| Female | 6 to 9 kHz | 3 to 6 dB |
| Rap with hard consonants | 5 to 8 kHz | 4 to 6 dB |
The right de-esser is never audible. What you hear is a vocal that stays clear even when it is loud.
When the problem is upstream
If you need more than 8 dB of reduction, the problem starts earlier: a microphone that is too bright, a singer too close, too much presence EQ. In the vocal compression chain you will see where the de-esser sits relative to the other stages.
Harsh or hissy vocal? Mix and master from $110, tuning included: the vocal comes out clear on every system.
Start a project