This falls into the category of compression but has a bit of an EQ twist to it. We are only compressing based on the energy in a narrow bit of the highs where esses (sibilance) are strong. This is typically between 4k and 8 kHz.

The best way to de-ess is prevention so that you won't need a de-esser. This is easy but probably not cheap. Use a great mic (hopefully tube condenser), a great mic pre (with minimal HF distortion), EQ higher frequencies or avoid boosting the 4K to 8K area and listen for problem sibilance as you are recording.

If the problem is a gap in the singers teeth rather than a gap in the budget, then a bit of cotton in the gap may help if the singer can tolerate it. If you have to use a de-esser, do it in the mix stage and patch it in after EQ and compressors.

Avoid de-essing the mix or forcing the mastering engineer to. It is much harder to de-ess if the vocal is surrounded by hi-hats, cymbals, snares, guitars etc.

It is easiest to use a dedicated de-esser and it hopefully does what you need. The goal is to reduce esses, not remove them. You should aim for "natural" dynamic reduction of the offending frequency band - not the esses total removal.

The THRESHOLD and RATIO depend on the song and the most natural effect.

The ATTACK can usually be slow (say "stop" and "top"- notice how slow "the ess in ‘stop’ starts). Usually between 70 and 200 ms is OK. Faster than this and the de-esser reacts to other consonants or other sounds.

The RELEASE can be a similar time constant. It is not to critical but again aim for "natural". Some de-essers compress the entire signal and some just reduce the highs only.

If you have to de-ess a mix - good luck. For mixes the ATTACK will be critical. You may get a dB or so of de-essing even with no esses so remember to use a bit of MAKE-UP to prevent overall loss of highs. If you want to live dangerously, you might de-ess more and boost the Make-Up Gain more, getting ever brighter and yet not making your ears bleed. Before you get excited, take a listen and judge whether it sounds like some kid or machine is playing around with the treble control of a stereo. Maybe too much, maybe magic.

A few last hints - Even more than with EQ, use the BYPASS SWITCH to check and verify you are doing only good and no harm. This will eliminate 8 out 10 headaches (and limiters needed in a session).

Spend more time learning the individual compressors and limiters with CDs that you are very familiar with. Play sounds into them and experiment - take notes and make comparisons. Getting the best stuff from most of them requires that you understand what the knobs do to music. Its not difficult. What is difficult, is understanding the language in the usual manuals and magazine features on compression. Learn your own units and plug-ins and translate what you have learned into terms that work for you. If in doubt go for "big", "smooth" and "natural" - these usually work.