MixMasterAI logoMixMasterAI
Sign in

De-esser

Reduce loud high-frequency sibilance with a linked high-band processor.

Up to 200 MB. Decoded at 48 kHz on this device. Long recordings may exceed the browser memory limit.

Use a vocal recording for the most targeted result. On a full mix, cymbals and other bright sounds are affected too.

How the de-esser works

What it does

The tool splits the audio at about 4.5 kHz into a low band and a high band. It follows the level of the high band with a fast detector (2 ms attack, 60 ms release). When that level goes above the threshold, only the high band is turned down. The low band, where most of the voice body sits, is never changed.

On stereo files the detector reads both channels and applies the same reduction to each, so the stereo image does not shift during an S sound.

Settings

  • Sibilance reduction: 0 to 100 percent. This sets the threshold, from -6 dBFS at the low end to -36 dBFS at 100 percent. A lower threshold means more S sounds are caught. At 0 percent nothing is changed.
  • The high band is never turned down by more than about 12 dB, so sibilance is softened, not removed.

Files and output

  • Mono or stereo files up to 200 MB. Files with more than two channels are refused.
  • Audio is decoded at 48 kHz on your device, so about 11 minutes of mono or 5.5 minutes of stereo fit in the memory limit.
  • The result is a 24-bit WAV at 48 kHz with the same number of channels. If the processed audio would go over full scale, the whole file is turned down by a fixed amount, and that amount is shown.

Good to know

  • Results are most precise on a solo vocal or vocal stem.
  • On a full mix, hi-hats, cymbals and bright synths also cross the threshold and get duller.
  • Too much reduction can make a voice sound lispy. Compare with the original at the same volume.

Adjust the threshold and compare before downloading.

How it works.

  1. 1Choose a recordingOpen a mono or stereo audio file. Nothing is uploaded.
  2. 2Adjust and listenSet the controls, process, and compare with the original. Keep only changes that help your recording.
  3. 3Save a new WAVDownload a 24-bit WAV with your plan access. Your original file is unchanged.

Questions.

What are the limits?

Use a vocal recording for the most targeted result. On a full mix, cymbals and other bright sounds are affected too.

Does it master my song?

No. This is a focused repair step, not loudness mastering. Only output that would clip is attenuated, and that amount is shown.

Can I use it on a full Suno song?

Yes, but cymbals and other bright sounds are reduced along with the S sounds. For the cleanest result, split out the vocal, de-ess it, and put the song back together.

What setting should I start with?

Start near 30 to 50 percent, process, and listen to the loudest S and T sounds. Raise the amount only if they still stand out.

Explore more.