I don’t remember ever using a de-esser in the early days of my recording career. Sure, there were a few clever engineers who would side-chain a compressor to tame certain frequency ranges, but that was rare and not always used on vocals. If a vocal was harsh or overly sibilant, we dealt with it through mic choice, mic placement, EQ, or coaching the singer. But honestly, it just didn’t seem like a widespread issue back then. I have a theory, backed by a bit of science, and no, it’s not that singers have somehow evolved to produce whistlier, sizzlier voices.
In the analog era, the gear itself was less prone to accentuating top-end harshness. Tube equipment, analog tape, early transistor designs, and even high-end mics of the day all tended to round off transients and roll off the extreme highs in a way that smoothed over sibilance. Even the harmonic distortion that came with the territory often worked in our favor. The frequency response and dynamic range of analog gear simply didn’t exaggerate those upper registers the way modern digital systems can.
Fast forward to now, and the de-esser is everywhere. It’s a standard feature on every input of most digital consoles. It shows up in every plugin bundle. I’ve always thought of EQ and compression as things that enhance a sound, tools that help you shape and present it. But the de-esser has always felt more corrective. It’s not there to elevate the signal. It’s there to fix something that’s already wrong.
So how did we get here? How did a tool that was barely used on albums that sold millions of copies and won shelves of Grammys become a standard part of every modern vocal chain?
At least part of the answer lies in the evolution of reproducing music digitally. With all its improvements, higher fidelity, extended frequency response, greater dynamic range, came a new byproduct: exposed harshness. The detail we gained came with a price. Sibilance that would have been softened or masked in an analog signal path now cuts through with painful clarity. And so, entire plugin bundles have been built to solve a problem that analog workflows rarely had in the first place, some even designed to emulate the exact smoothing effect that analog gear used to give us for free. Consider all the digital versions of vintage compressors, tube levelers, EQs, and analog tape emulators.
What’s interesting is that there’s a whole generation of engineers now who have never heard analog multitrack tape in the real world, much less had to edit it with a razor blade. For them, the de-esser isn’t a fix. It’s just part of the chain. And in a way, it makes sense. When the tools themselves introduce problems that older systems never had, of course the fix becomes standard.
I’m not saying all this to be the old guy yelling at young engineers to get off my analog lawn. Using the de-esser has, in some ways, spoiled me, or at least spoiled my ears. I’ve noticed that I’m now so accustomed to smooth, polished vocals and spoken word, that when I hear content without a de-esser, especially one that clearly needs it, I find myself cringing. It’s usually a podcast or a livestream, and the “S” sounds jump out like static on a phone call. I think, “How did the person producing this not hear that?”
The answer is simple: most people don’t hear it the way we do.
As audio engineers, we get hyper-aware of certain things. Sibilance is one of them. We hear what others don’t, and after a while, our ears start expecting and craving a certain level of polish. But that doesn’t mean everyone else is bothered by what we notice. I’ve had plenty of conversations with worship leaders, music directors, and even other techs where harsh vocals never came up. I’ve pointed out what sounded like painfully sharp sibilance to me, and nobody else flinched. That tells me not to overthink it. Not every voice needs a de-esser, and not every mix needs to be smoothed to perfection.
In the real world of live mixing, sometimes we’re changing the oil in the bus while driving down the road. Maybe rehearsal got canceled. Maybe the virtual soundcheck computer ate the tape. Maybe it’s a new guest vocalist on a mic we haven’t had time to dial in.
In all the years I’ve had to build a mix on the fly, I’ve never actually written down how I triage a mix. But I know this: when time is tight, in the interest of not letting the pursuit of perfection interfere with quickly achieving very good, I’m not reaching for the de-esser first.
I’d rather get the vocals blended, compress what needs control, and clean up the EQ on anything that might ruin the mix. If I don’t make it to the de-esser, that’s fine. The mix can still be excellent.
That said, when I do engage a de-esser, or when I’m using a dynamic EQ or multiband compressor to do the job, I’ve picked up a few nuggets over the years that might help someone else:
Vocal sibilance is usually lower than you think.
A lot of people start looking for sibilance up in the 10kHz range. But up there, you’re mostly hearing air and sparkle. The actual sibilance—the part that feels like it’s spiking, often lives between 5kHz and 8kHz. That range depends on the voice and the mic, but most of the trouble isn’t happening in the top shelf.
Finding the problem without causing another one.
If you’re leisurely mixing an album, you can take all the time you need to zero in on the exact frequency contributing to a vocalist’s sibilance. But in a broadcast or FOH setting, especially under time pressure, even virtual soundcheck has its limits. You may not have that luxury.
Instead of chasing surgical precision right away, start slightly below where you think the problem is. Set the threshold so it’s taking out too much on purpose, then slowly move up until the clarity returns. This often leads to a more natural and transparent result faster.
If you’re soloing or cueing the vocal to experiment, that’s fine. Just remember to always listen in context before moving on. That’s good advice for any part of your mix. It’s easy to overdo something in solo mode that doesn’t translate well in the overall mix.
Use the RTA as a starting point, not as gospel.
Most digital consoles feature a per-channel RTA display, which is a huge help. Watch the display as the vocalist hits their strong sibilant sounds. It’s usually easy to spot where the energy is building up. That can help you find your frequency starting point. Just keep in mind that the RTA shows the input, not necessarily what’s coming out of the PA speakers or your studio monitors. It’ll be close, but there’s a lot of processing between the channel input and what the audience actually hears. If you’ve got a Smaart rig with a reference mic showing you an RTA of the actual room, even better. That’ll give you a picture of what’s really hitting the room or the stream. Either way, the RTA is a guide, not a final answer.
Presets are rarely set-it-and-forget-it solutions.
There are a number of de-esser plugins on the market, and many include presets labeled things like Female Vocal, Proximity, Broadcast, or Air Control. Oh, that it were so simple! These can be helpful starting points, but they almost always require manual adjustment, customized to your mics, your singers, and your system.
This is true for virtually any plugin. If the Female Vocal preset happens to work perfectly for one of your singers, great. But don’t count on it working across the board. Different voices, microphones, and delivery styles all affect how sibilance behaves, and no preset can account for that.
The point of all this isn’t to ignore de-essers.
I use them regularly. But I try not to slap one on just because that’s what we’re supposed to do. And if you haven’t practiced a little off-line, you can overdo it, where it does more harm than good.
As a legacy habit from my analog days, I don’t like to clutter a signal chain with unused (bypassed) plugins. In the digital realm, does a plugin in bypass mode degrade the signal or add latency? The answer is probably yes, but it’s debatable whether those artifacts are actually audible. When in doubt, and if you know you’re not going to use something in the chain, just take it out.
Some vocalists need it. Some don’t. If it’s a time crunch and I have to decide between working out a smooth vocal blend and zeroing in on the exact sibilant frequency on a few of the singers, then nobody needs it. Some days, you’ve got 99 problems, and the de-esser isn’t one of them.
Still, I’m glad we have the tool. It’s good to have around when it’s needed. We should remember that not everything needs fixing. Sometimes the best thing we can do for a vocal is back off and let it be human. What’s your experience with de-essers? Maybe you’ve never touched one, and that’s okay, as long as you don’t forget to listen.
