Remove Harsh Sibilance from Vocals
Pick a vocal, narration or podcast file and dial in how hard the esses get ducked — de-essing runs in your browser, nothing is uploaded.
About this tool
A de-esser reduces the short, harsh bursts that make s, sh, t and ts sounds jump out of a vocal, narration or podcast track. It is different from a normal EQ cut: the high band is ducked dynamically only while sibilance is present, so quiet breaths and the general brightness of the voice stay closer to the original.
This tool uses ffmpeg's dedicated deesser filter. The controls are exposed as honest 1-100 scales because the filter's amount, crossover and reduction controls are unitless coefficients, not fixed Hz or dB values. Start with the default vocal preset, then use ess mode to listen only to what is being removed; if you hear too much of the voice body in that audition track, raise the band or lower the amount.
Example settings for a vocal WAV:
gizza tool de-esser 'url=https://example.com/vocal.wav' amount=60 band=70 max_reduction=50 mode=output format=mp3
For a gentler podcast polish, try amount 35, band 80 and max reduction 40. For troubleshooting, switch mode to ess and format to wav; the output should mostly contain the harsh esses, not the whole voice.
Limits and edge cases: this is not broadband noise removal, a click remover or a full multiband mastering chain. Very aggressive settings can create a lisp or dull consonants. Embedded album art/video streams are dropped because the audio is re-encoded.
FAQ
Is this the same as lowering 6 kHz with an equalizer?
No. A static EQ cut lowers that band for the whole file, which can make the entire vocal dull. This de-esser ducks the upper band dynamically only when sibilance is detected. Use audio-eq when you really want a constant tone change.
What should I listen for in ess mode?
Ess mode renders only the material being removed. Ideally you hear mostly sharp consonants and very little vowel or voice body. If the audition output sounds like the full vocal, raise band, lower amount, or lower max_reduction.
Why are the band controls not labeled in Hz?
ffmpeg's deesser filter uses unitless coefficients whose exact crossover depends on the source sample rate. A fixed Hz label would be misleading, so the page uses 1-100 controls and describes the practical direction: higher band values keep the effect closer to the very top of the spectrum; lower values treat more of the voice as sibilance.
Which output format should I choose?
Use MP3 for sharing, WAV or FLAC when you plan to keep editing, OGG for web projects, and M4A when you need an AAC-style file. De-essing rewrites the samples, so every format is newly encoded.
Developer & Automation Access
Run it from the terminal
Same engine as this page, headless — via the gizza CLI:
gizza tool de-esser 'url=https://example.com/input' 'amount=60' 'band=70' 'max_reduction=50' 'mode=output' 'format=mp3'New to the CLI? Get gizza →
Open it by URL
Pre-fill and auto-run this tool with query parameters — the names match the API/CLI:
https://gizza.ai/tools/de-esser/?url=https://example.com/input&amount=60&band=70&max_reduction=50&mode=output&format=mp3Machine-readable descriptor: tool.json — title + parameters JSON Schema for agents.