Tune a de-esser by listening to three consonants

Record a short line such as “She saves the last seat,” followed by a sustained vowel and a quiet breath. A de-esser turns audio down when its detector identifies excess energy in a selected range. The task is to tame the word that leaps forward while keeping the words easy to understand.

Build a test the meter can fail

Choose three moments from a real take. Mark a piercing S, a softer SH that belongs to the performance, and a bright vowel you want untouched. Include a few seconds around them so you can hear the words in context. Save a dry reference. Put a de-esser on a copy at the same level, with the rest of the mix available on demand. Begin at a comfortable listening volume, especially when monitoring a narrow band of high frequencies.

Use the detector’s listen or filter-solo mode if your processor offers it. Move its frequency control until you recognize the piercing consonant distinctly. Listen for whether the quiet vowel also dominates the detector. Exit solo and adjust the sensitivity or threshold until the strongest S triggers reduction. Limit the maximum reduction and increase it only until that consonant sits alongside the rest of the line. Apple’s DeEsser 2 has separate controls for the monitored filter, threshold and maximum reduction. Apple’s DeEsser 2 controls

Compare words, then the mix

Alternate between the dry and processed versions on the same phrase. Listen first for the difference between “she” and “see,” then for whether the vowel changes color. A lisp or a dull vowel calls for less reduction, a narrower detection range where adjustable, or a different frequency. If the processor misses a single unusually loud S, a small gain edit to that consonant can solve it while leaving the remaining line alone.

Unmute the arrangement. A cymbal might be supplying the hiss you attributed to the voice. Mute the vocal for one pass to check. If compression later in the chain brings the problem back, compare de-essing before and after that compressor at matched vocal level. The better position is the one that controls the marked S and preserves the quieter SH in the finished mix. Print ten seconds of both options and label the preferred one with the offending word and the processor position. That note makes a second listening pass far more specific than “vocal too bright.”

Companion chapter · Mix a song around the part the listener follows

Bedroom to Stage contents and glossary

Back to blog

Leave a comment