Diss the Hiss


According to the Merriam-Webster dictionary, “sibilant” (the adjective form of sibilance) is defined as “having, containing or producing the sound of or a sound resembling that of the s or the sh in sash.” More telling, perhaps, is the contextual example the dictionary provides: “a sibilant snake.” This emphasizes the strength and importance of the ssss sound. Sibilance, as it is referred to in recording circles, is a collective of consonant and combination sounds that tend to show up harshly in vocal recordings, including s, sh, c and ch sounds. On the other hand, de-essing is, as the name applies, the act of removing or lowering those sibilant sounds from a recording. Knowing what causes sibilance and how to de-ess will shine some light on how to take your basement and bedroom vocal recordings to a more polished level.


Sibilance is a natural phenomenon of spoken language. The particular collective of sibilant sounds are crucial to many words and phrases in just about any language (though they show up more in some than others). The problem is, when it comes to musical vocal recordings (or even simple spoken-word), the sibilant sounds often show up many decibels louder than most of the other sounds and can be quite harsh to the ears if they aren't corrected. The problem rears its ugly head to a greater or lesser degree depending upon the specific singer on hand, and it can be enhanced (or reduced) in a number of ways related to the recording process, including mic placement and proximity, the sound of the recording room or space, applied compression and the engineer's choice of microphones. Because each recording situation — not to mention each singer's voice and studio experience — is different, there are no hard and fast rules to speak of. But for starters, follow these basic tenets: If you have choices, be very selective with your mics when recording. Or when purchasing a mic for a specific singer and project, find one that does not overly emphasize the sibilant sounds, and use your own ears as the judge. Then apply that same care to how you place the mic for the recording sessions — if you are in a highly reflective space, avoid placement too close to walls or right in the middle of the room, and experiment with the distance from the vocalist's mouth — close enough for clarity and strong overall presence, far enough to avoid the snake syndrome. If you wish to use compression, try using it for the express purpose of de-essing, which is covered below.


The value in de-essing is in yielding a more even-toned, professional-sounding recording, with much less (or ultimately none) of the harshness of overly emphasized sss, sh and ch sounds. De-essing can be achieved in a number of similar, yet different ways. The most common is sidechained, frequency-dependent compression, whereby an equalizer is inserted into a compressor's sidechain, and a chosen band of frequencies that spans the sibilant sounds is boosted by several decibels. Because the boosted “esses” will trigger the compressor more strongly, the compressor will reduce the overall signal at precisely those harsh points. However, because the entire signal will be compressed, an increased emphasis is placed on the compressor's attack and release times if you want this to succeed without “pumping” artifacts. Start with a very fast attack time and fairly fast release time, and slow them down only as necessary. Again, use your ears as the judge. (For more on sidechaining, read the Phantom Power article “Do the Dip” in the August 2007 issue of Remix or on remixmag.com.)

A cruder but still useful way to de-ess is to process the vocal with just an EQ, using very selective notching-out of the troublesome sibilant frequency bands. That will color the resulting recording more, since the notch will affect more than just the sibilant portions, but if you don't have access to a compressor with a sidechain, this may be a viable alternative. If you are fortunate enough to own a dynamic parametric EQ, your problem may be more simply solved as you zero in on the sibilant frequencies. When the vocalist belts those out, the EQ responds in kind by reducing those frequencies.

For another technique, use multiband compression; thankfully, more and more DAWs today, such as Cakewalk Sonar 7 Producer Edition ($619; www.cakewalk.com) and the new MOTU Digital Performer 6 ($795; www.motu.com), include these highly useful tools in the standard bundle. Just as with the EQ in the sidechain description earlier, the idea here is to determine the frequency range of the sibilance and more heavily compress just that range, while leaving the other bands either unprocessed or much more lightly compressed. Even a simple 2-band compressor could do the trick, but it may prove even better if you can pass through frequencies both above and below the sibilant range with a 3-band (or more) compressor.

As a third option, apply some good old-fashioned elbow grease in your DAW. Use your automation curves and draw in explicit drops in volume wherever troublesome sibilance occurs. The downside of course is the time required to do that very effectively, yet the upside is the opposite: that surgeon's level of control that you have to sculpt the perfect vocal track. For some engineers these days, this level of micro-tweaking has become standard fare, just like tape splicing became in the pre-Pro Tools, Cro-Magnon recording days.

The above techniques are not necessarily just for minding your “esses” and “cees.” You can use them to tame long, drawn-out screams; Ps, Bs, and Ts; and other strong-attack consonants and even sounds unrelated to vocals, such as overly boomy bass, saxophone squelches, VU-pinning trumpet blasts and many other unruly hot volume troublemakers. Try out these tips, but as always, practice makes perfect.