Pro Audio Files

Tips for Controlling Vocal Sibilance

Vocal sibilance is an unpleasant tonal harshness that can happen during consonant syllables (like S, T, and Z), caused by disproportionate audio dynamics in upper midrange frequencies. Sibilance is often centered between 5kHz to 8kHz, but can occur well above that frequency range.

This problem is usually caused by the actual vocal formant, but can also be exaggerated by microphone placement and technique. This article will discuss some ways to control vocal sibilance, and keep the problem from becoming a musical distraction.

Sibilance at the Source (best read with sibilant whistle)

In phonetic terms, sibilance comes from a type of vocal formant called a fricative consonant. During these sorts of utterances, the airway (usually the mouth) is drastically constricted by two anatomical features, like the teeth, tongue, or palette.

This pressurization causes some amount of noise that forms the consonant sounds we would recognize from a phase like, “Sally sits sideways on the tennis trolley.” Sibilance is a very necessary feature of human speech, but when there’s (subjectively) too much noise created during these consonants, we get a very distracting harshness.

It isn’t really practical or productive to address micro-muscular vocal technique during a session, so your best bet to mitigate sibilance at the source is microphone selection and placement. Here are a few suggestions:

  • Every vocalist is remarkably different, so don’t pre-suppose that anything you’ve tried before will or will not work again.
  • Be sure to leave some space between your vocalist and the microphone. Twelve to eighteen inches would be a nice starting point.
  • A pop filter won’t do anything to help with sibilance.
  • Once you find a microphone and distance combination that helps, try angling the microphone downward 10 to 15 degrees to place the 0-degree axis toward the throat instead of the sibilant source. 

Audio Dynamics Processing

Vocal sibilance is a phenomenon of disproportionate dynamics within an isolated frequency range. In other words, it is a problem of too much loudness contrast within a small frequency range of a waveform that has a dynamic profile of its own.

‘De-essing’ is the classic compressor technique used to address vocal sibilance through processing. In fact, de-essing is just one example of many uses for compression that is conditioned on a limited frequency band, or a modified harmonic profile.

De-esser Signal Flow

Audio dynamics processors like compressors and expanders contain two signal paths:

  1. The audio path, which is subject to conditional gain reduction and;
  2. The sidechain or ‘key’ path, which the gain reduction is conditioned on.

In short, gain reduction happens (or not) in the audio path based on the interaction between the sidechain signal and the detector settings (i.e. threshold and time constants). By placing an EQ in the sidechain path, we can further condition gain reduction on user definable frequency conditions.

The de-esser technique typically uses a narrow peak EQ in the sidechain path to boost the most offensive sibilant frequencies. This EQ exaggerates the dynamic difference between the sibilant band and the rest of the vocal waveform, making it much easier to achieve gain reduction during those consonants (and only then).

A pre-configured de-esser may provide an interface as simple as a compressor threshold and the peak EQ center frequency. These often work just fine. For more detailed control, one could patch an EQ into the sidechain of a relatively fast compressor, or use any number of compressor plug-ins that provide detailed EQ in the sidechain path.

There are lots of great techniques based on this signal flow, so spend some time with it. Frankly, de-essing is the least of what you can do by adding frequency conditions to your gain reduction.

Other Precautions

When you’re recording a vocal performance that may have a sibilance problem, resist the urge to compress the signal in the channel path. Over-compression can exaggerate sibilance. Instead, try using a fader to level the vocal performance, or just record with an adequate amount of headroom.

The same applies to the mixing process. Once you’ve done your best to control vocal sibilance, try using a fader and automation to maintain a consistent vocal volume in the mix. If you simply must instantiate a compressor on every vocal track, keep the attack time slow (> 30ms), and the ratio low.

Finally, don’t listen too loudly when you mix. That’s good general advice, but quality control issues like sibilance highlight its importance. Try a control room volume of 78-83dB(C) SPL. You might be surprised how much detail you’re suddenly able to hear.

Missing our best stuff?

Sign up to be the first to learn about new tutorials, sales, giveaways and more.

We will never spam you. Unsubscribe at any time. Powered by ConvertKit
Rob Schlette

Rob Schlette

Rob Schlette is chief mastering engineer and owner of Anthem Mastering, in St. Louis, MO. Anthem Mastering provides trusted specialized mastering services to music clients all over the world.
Smiley face
  • I’m glad Rob wrote about this. I agree that the BEST approach is to choose a different mic and/or mic position. Once you’ve recorded excess sibilance, all other solutions are compromises.

    I’ve never had good results pointing away from the mouth, so I wouldn’t suggest pointing at the neck. But I would recommend moving the mic back, higher, to the side, or even lower so that the relationship to the mouth is changed (assuming a directional mic is being used).

    Darker microphones (like the AT4047) and mid-range focused microphones (such as the Lawson L47) can really make a difference, but it all depends on how a mic and a person’s sibilance interact. As Rob suggests, there is no single solution.

    In my experience, male sibilance tends to be in the 5-7k Hz range and female sibilance in the 6-9k range. In rare cases I’ve tuned a de-esser as low as 3k Hz and as high as 11k Hz.

    Some of the better de-essers give you the option to only compress above a set frequency, or compress only a frequency band. These tend to sound more natural, in my opinion.

    One of my favorite de-essers is analog: The DerrEsser by Empirical Labs (originally found in the LilFreq). I recently tried the McDSP DE555 which sounded better than my standby Waves Renaissance DeEsser. Sometimes a multi-band compressor that’s only active in the sibilant range works great too.

    Frequency editing can be done with software like iZotope RX or Frequency View in Wavelab. This is actually one of the best sounding ways to tame sibilance post-recording session, but it is far more labor intensive than “set and forget” de-essers.

    • Rob

      Hi Randy. That EL DerrEsser was really amazing for me on a record I mixed filled with really exposed, not-so-well recorded acoustic guitar. I haven’t had the opportunity to use one again, but I was very happy with it.

      Thanks for reading.

    • neinonein

      Rob, how would you go about adding sibilance or ESSes in case of bad vocal processing (low ESSes) Thanks.

    • I would get out a mic and re-record. You can play with upward expansion, but the frequency condition makes that a long shot.

    • A.T. Rips

      There is a guy that I know that knows how to remove vocals without distortion but he does not want to tell me how he does it. He gave some hints that spectral frequency display is required and it is something that audio engineers know how to do. Do you know the precise steps of how it is possible? I suspect it is related to this.

  • Sibilance is a serious pain in the *** and for me is probably what I spend the most time on when mixing vocals. Like Randy I recently got the mcdsp DE555 and along with sonnox supressor are my go to plugin de-essers. I have not used the deresser but have thought about it a few times. But the more I think of getting one the more I realize I probably need to fix the problem with mic or placement first. Great post Rob

    • Rob

      Thanks, glad you liked it.

      I really like the Sonnox de-esser too. It has really flexible EQ parameters, making it much easier to zero in on the offending sub-band.

  • Great information, and well written. Very helpful. Angle the mic? That is what I try to do, and it works!

  • [Some additional tips on controlling sibilance from a previous article I wrote]

    At The Source:

    1. Choose the right mic!

    This almost goes without saying, but if the vocalist is overly bright, you might want a darker microphone. A ribbon mic, a softer dynamic mic (like an RE20), or a vintage sounding condenser (U67), might be a good grab. Something that has a rolled off, smooth top end, that will take well to EQ. There is no con to this approach.

    2. Choose the correct mic position.

    You may want to tip the mic a bit off axis, aiming it slightly left or right of the mouth, or perhaps somewhat down. Angling the mic will mostly change the way the treble range is picked up, as lower frequencies are somewhat less directional. The con is that the evenness of the frequency response will be somewhat disrupted by grabbing an off-axis response.

    3. The bubble gum trick?

    If sound projecting from the upper palette is the problem…. eliminate the upper palette. Have the vocalist chew up some gum and stick it to the roof of his/her mouth. This will cut down the sibilance significantly. The big con here is that this can be awkward for a performer – if it throws off the performance or sense of pitch, it’s not worth it. But it’s an option.

    In The Mix:

    1. Manually ride the fader.

    Hear an “s”, turn it down. This is a transparent approach. The con is that this is time consuming.

    2. Wideband De-Ess

    Another basic approach, this is compression that is reacting only to the frequency range. This is much faster than fader riding, however, it tends to leave the leading edge of the “s” unaltered. It makes your sibilant sound less intrusive but spikey, and may be just as annoying. The other con is that you’ll tend to catch some of the treble of non-sibilant words and pull down the overall “spark” of the vocal.

    3. Frequency selective De-Ess

    This is the same as wideband, except instead of turning the whole signal down, you’re just turning down the treble range, as opposed to the whole of the signal when it triggers. This is good for evening out the tone, but has all the drawbacks of wideband, plus it induces eq artifacts (although they are fairly minimal).

    Here’s some that can get better results, although they require a bit of trickery. I’ll start with my favorite.

    1. Pre-triggering the De-esser.

    This is a really cool technique. It’s a major pain in the ass to set up in analogue mixing, but it’s easy in digital. Make a copy of your vocal on a separate track. Move the copy ahead of the main vocal by 50ms. Put a De-Esser, or a multi-band compressor (like Waves C6) – something with an external sidechain – on your main vocal. Key it to the copy.

    Through this setup, every time an “s” comes through on the copied signal, it will activate the de-esser on the main signal – but it will do it about 50ms earlier than when the actual “s” from the main signal would occur. This allows the de-esser to reach peak gain reduction BEFORE the “s”. If you set the release for about 100ms, you’ll knock out that “s” sound without leaving any spikes on the leading edge. This is a very transparent way to do this.

    2. Over De-Essing.

    Another way to get rid of “s”s is to go overboard with the de-esser, so that it’s working even on parts of the vocal that aren’t sibilant. If you then feed this into an EQ, you can boost your high end back up to regain your lost treble. The pro of this is that it has the added benefit of making the treble of the vocal very present without making the “s”s jump. The con is that the high end will almost assuredly become less smooth unless you are using a really good de-esser and treble boost. I use this technique on vocals that only become sibilant once I add a lot of high end to them – which is fairly common in Pop/Hip-Hop/Dance productions.

    3. Smooshing your high end.

    This one takes a little guts. One of the beautiful properties of minimal phase EQ, is that if you use the exact same EQ and do the exact same amount of boost, followed by the exact same amount of cut, you will nullify your artifacts and come out with the same signal as when you started. Using this principal, you can add like 10-20db of treble gain to your vocals, compress the vocals, and then do the same amount of attenuation after the compressor and get a surprisingly transparent form of de-essing. It sounds weird because it’s so extreme, but you get your normal compressor reacting more to the sibilance. Give it a try, you may be surprised. The major benefit to this is that it works well in the analogue realm, not just the digital.

    There are other techniques for easing out sibilance. These are the ones I’ve found most useful. I should also say that using a combination of these techniques will probably get you further with less artifacts.

    • Courtez Presberry

      Do you have any video on this Mr. Weiss?

  • what a pain sibilance has been for me over the years! even on my own voice.

    as others have said choose the right mic. i am using dynamic more and more these days. i really think home studio people should take a second look at dynamics.

    but after the damage is done i rely on 3 things.

    1) cutting some high shelf eq. not too much that it sounds funny.

    2) then using a multiband compressor with crossover set to about 1k, wide q, and ratio of 4:1. again adjust so it sounds “natural” (totally meaningless term)

    i almost always use eq and multiband together. a bit of each instead of tons of one.

    you can also slice out the sibilance into short chunks and lower their levels. make sure the fades are long enough for this to sound….er….natural. i think most daws can do that these days.

  • Some sound (no pun intended!) advice here.

    My voice always comes out incredibly sibilant on my recordings and it took me years to find a mic that could deal with that. Apart from a few boutique tube mics I have used at other studios, which I can’t afford myself, the one I found that works for me is the Blue Baby Bottle. It’s aquite a dark mic and I also use the angled-down trick too which helps.

    There is a handy trick I use, which is effectively the same as the fider riding trick during tracking. If you adjust the trim of the recorded audio file on the sibilant sounds then you can tame the level going into any subsequent compressor during mixdown. This is a little time consuming until you get used to doing it a lot, but it is very effective in getting a smooth vocal that still pushes through the mix.

    Nice article thanks for the insight and the tips.

    Rob Cooke.

  • Knight

    thanks alot for the info, sibilance is horrible for me, I got a mic that I really like (AT 40/40) but the sibilance is still there, im using pro tools, and becuz de-essers barley do much I usually go in and manually bring down the volume of the hyped S’s, but its a pain to do that. I heard about mic placement being able to help alot so im definitely gonna try this method out, thanks. I was just curious tho, how much is 15 degrees? I dont want to over do it im sure, but im assuming u mean by tilting it slightly

  • jimfoon

    A lot of great suggestions here! Just my 2 cents from the post editing perspective (Of course always try to get the best recording possible first!!!). Once in a while I’ll use a tape saturation at the end of my chain IF its content appropriate and use a 15 or 7.5 IPS setting and kiss the red a little to really round things out. Slate Tape Emulation or REEL Tape Saturation, which has some nice settings for SWISS/USA/LO-FI style tape machines SWISS being the brightest and LO-FI Being the darkest/dullest. Of course the slower speeds of the tape used also changes the overall sound and typically anything under 30 IPS starts to roll high end which can be a nice thing. Definitely an alternate approach but has worked great for me in certain situations!

    If trying to get a super clean sound I also would rather use an EQ if I were going to manually go in and only gain reduce the frequencies that are harshest rather than the whole clip of the spike. Audio Suite in Pro Tools or any rendered plugin makes it easy to select a little more than you need and then pull back the sides to get just the right moment of sibilant spike out. Same trick works for plosives and low end frequencies by using a high pass filter on offensive words.

    Favorite De-Esser by far is Fabfilter Pro DS. Has great auditioning features, excellent meter/waveform display, supports external side chaining, tons of modes and styles, has advanced monitoring and audition functions, and best of all a Look Ahead Feature which is super important in squashing those peaks! I Sometimes use 2 in different freq ranges. Don’t forget to tick off the 4x oversampling option at the very bottom for highest quality engine too.