De-Esser

Sibilance sits roughly between 5 and 10 kHz, and it gets worse with close miking, bright condensers and any broad high-frequency boost in the chain. A de-esser listens to that band and ducks it for the fraction of a second the S lasts. The rest of the voice is untouched, which is what a plain EQ cut cannot manage without dulling everything else along with it.

Set the frequency by ear first, usually by soloing the detection band, then set the threshold so the meter only moves on the worst syllables. Overdo it and you get a lisp. That sounds considerably worse than the sibilance you started with.