De-esser
A compressor that only listens for "s" sounds, and tames them before they slice into the listener's ear.
Sibilants — s, sh, ch — carry a lot of energy between roughly 5 and 10 kHz and turn harsh easily, especially with a bright microphone close to the mouth and a compressor lifting everything quiet. A plain EQ cut in that region would dull the whole voice. A de-esser cuts only while the sibilant is actually sounding.
Technically it is frequency-dependent compression: the detector signal is filtered so the compressor reacts only to the treble range, and the gain reduction is applied either to the whole signal or just to that band. Analogue versions were already in radio and film studios in the 1960s. Set too hard, it gives you a singer with a lisp — a fault as common as the one it was meant to fix.