Good point. However, most de-esser I use in the studio (from Waves and the stock Pro Tools de-esser) actually have an option where the de-esser ONLY affects the high frequencies, everything above the threshold.

You're right that a de-esser can affect the entire frequency spectrum of the track, but I thought I'd add a little clarification. Thanks!

Joe

A statement is made that a de-esser compresses a selected frequency band, and only that range of frequencies. This is not true. Although there are some specialized multi-band compressors out there, the standard de-esser that is included in most DAW programs is a full range compressor.

The difference is that the de-esser takes advantage of what is known as a "side chain", which is simply a secondary signal route that sends the original signal (usually a vocal) through the equalizer and then on to the signal level sensing part (circuitry or computer code) to trigger the gain reduction. The main signal route through the de-esser remains un-equalized - highs are not boosted to the signal you hear, they are only boosted for the benefit of the processor's level sensing, so that it is more likely to react to high frequency content and less likely to react to anything else. In other words, the de-esser's side chain equalization "stacks the deck" to change how the compressor will react to the incoming signal.

Why is this rather picky distinction important? You should remember that applying a de-esser to a vocal track will affect the whole track, not just the highs, unless you take care to insure that the various settings - the side-chain EQ, the threshold, and especially the attack and release times - are set just right. If the EQ is set up improperly, or if the attack/release timings are not just right, you could wind up with a track that sounds choppy and uneven, while still having a sibilance problem.

The reason that these processors work at all is that they take advantage of the naturally occurring phenomenon that is a characteristic of most human speech: sibilant sounds are rarely, if ever, accompanied by any other vocal sound - your vocal cords do not vibrate at the same time as you make an "s" sound. So applying a quick, "in-and-out" gain reduction to the whole audio stream, but just when the sibilance occurs, has the net effect of calming down the jarring and intrusive "s" and "f" sounds, while leaving everything else alone.

Some people are just more likely to produce problematic sibilance than others. You can often determine this by simply listening to them talk. Differences in mouth and jaw size, along with how the teeth and tongue interact, can result in some people having a constant, overwhelming amount of sibilance in their speech and singing, while other people produce very little (what's called a "lisp").

To further complicate matters, you have the different frequency responses of various microphones, the desire to add "air" or a high frequency EQ boost for clarity, as well as the phenomenon described in the video where overall compression can have the effect of making quieter sounds more prominent, thus intensifying problematic sibilance. Where you place your de-esser in your signal chain can have a huge effect on whether or not it is able to fix the problem you've identified.

Sometimes, other methods should be explored when you know that you will be dealing with someone who is likely to have a greater issue with sibilance. Choosing a microphone with a smoother or even reduced high frequency response might be the best starting point. Placing a de-esser after you've applied a lot of high frequency EQ boost, and/or a lot of compression, might actually have less effect than you might want, due to the de-esser's sensing circuitry/code being less able to distinguish between actually voiced sibilants, and a microphone's peaky high frequency response coupled with an "air" boost from some EQ, with less contrast between soft and loud sounds resulting from aggressive vocal compression. You might be better off putting the de-esser at the front of the signal chain, or in the middle. In extreme cases, you might want to de-ess before AND after the rest of the processing chain.

Just remember that you need to pay careful attention to what the de-esser is effecting. Watch the meters to make sure that the gain reduction is only taking place when you need it to, and not having a negative impact on the rest of your vocal track - because a de-esser, despite its specialized side-chain equalizer, is a full range device.