Great sounding recordings through skills and knowledge without expensive gear
Check out our YouTube Channel
Join the FaceBook Community
Follow us on instagram
Advanced De-Essing
Fix that lisp, get natural sounding S's even on the most stubborn recordings.
De-essing is mostly used for vocals, but let’s not limit ourselves to just that. Once you understand how it works, you can use the plug-ins creatively for a wide variety of applications.
This is an advanced lesson which assumes you already know basic de-easing. In this lesson we will cover
1- what causes sibilance.
2- identifying sibilance frequencies
3- strategies for reducing the specific frequencies.
Part 1- what causes sibilance
To understand how to eliminate sibilance, first we need to understand what causes it. The original recording might sound great, but during the mix for some reason the sibilance gets out of control. There’s 2 things working together to cause this. EQ and Compression.
Quite often during a mix, the treble will be brought up on vocals. For the most part, this sounds great, except that eq boost will make the sibilance more prominent.
And then, to add insult to injury, it gets worse by compression. A compressor will reduce the volume of the loud parts, and increase the volume of quiet parts. It senses volume by the amount of energy in the incoming sound wave, which is not the same way our ears hear volume. Our ears will hear high frequencies with much smaller wave amplitude at the same volume as lower frequencies with a much larger wave amplitude. Since lower frequencies have more energy, the compressor will react to them much more. When we pronounce an “S” or a “T” these are high frequency sounds. There is very little energy in themselves, and as far as the compressor can tell, this sound is low volume. So, the compressor does its job and makes it louder.
Let’s try it out.
She sells sea shells and sibilance sucks
Here it is with a bit of a treble boost.
Here it is with just a compressor
And here it is with both.
Part 2- identifying the frequencies.
The more prominent frequencies that are heard with sibilance are in the range of 7-9 kHz. The way a de-esser works is it listens for those frequencies. When a vocalist is pronouncing vowels, there’s very little volume in those frequencies, then when an “S” or a “T” is pronounced, there’s a sudden surge in volume in that frequency. When there’s high frequency sounds that surpass a set threshold, the de-esser kicks in, and is basically a compressor that only reduces the volume of the those frequencies.
Depending on the vocalist and the microphone used, the peak frequency of the sibilance could be anywhere between 7 kHz to 10 kHz. So we find the peak frequency, reduce it, and then what happens... it sounds like the vocalist has a lisp. Sometimes you’ll encounter a particularly difficult vocal where there’s no in between, it either has too much sibilance, or sounds like a lisp. The good news is, this is possible to fix, and make it sound natural.
When we de-essed and it sounds like a lisp, the goal is to make that lisp sound natural again.
What causes the lisp is that the high sibilance frequencies were reduced in volume more than the lower sibilance frequencies. There’s 3 frequency ranges of sibilance. Low, mid and high. A natural sounding S will have a nice balance of all 3.
So let’s break sibilance into 3 frequencies ranges
1- low frequency sibilance. 4K- 7k.
2- mid frequency sibilance. 7k- 9k
3- high frequency sibilance 9k- 12k
1- low frequency sibilance. 4K- 7k
This is the meat of the sibilance. If this is disproportionately loud, the S will sound like a lisp.
To correct the lisp, you can make this frequency range quieter, or make the higher frequency ranges louder with a regular eq.
2- Mid frequency sibilance 7k-9k
This is usually the most prominent frequency range of sibilance, and if you’re just going to focus on one area for light de-essing, this is it.
3- High frequency sibilance, 9k-12k
This area sounds really harsh, trashy, and almost percussive when it’s too loud. The sibilance will sound like a lisp if this range is too quiet.
When performing de-essing, especially heavy de-essing, you need to listen carefully to the balance between these 3 frequency ranges and reduce their volume appropriately. To get a natural sounding S, all three of these need to each be at just the right volume.
Part 3- Reducing the specific frequencies.
Sometimes you just need light de-essing, and a single de-esser plug-in is adequate. But sometimes you need heavy de-essing, and this can be challenging to make it sound natural.
I have 3 techniques which I use, and sometimes, I use all 3 in combination with each other. Using all 3 works well because too much of one thing won’t sound natural, so a little bit of several different things works better.
1- use 2 de-easers one after another
2- side chain a compressor or a dynamic EQ.
3- automate quick volume dips, or compressor settings.
1- use 2 de-essers one after another.
Try this first. If 1 de-esser plug-in sounds good, but not quite perfect, listen closely to the sibilance, and which of the 3 frequency ranges is too much. Use another de-esser to target those frequencies.
2- Side Chain a compressor or a dynamic EQ
Use this in combination with the first strategy. This is a very effective for heavy de-essing. There’s no limit to how much reduction you can perform on just the sibilance. You can render all the S’s silent if you wanted to.
This technique is not just a plug-in, it requires a little bit of setup.
Duplicate the vocal track which you wan to de-ess.
On the duplicate track, remove all plugins and aux sends. Insert an EQ with a steep slope low cut filter at 7 KHz, and high pass filter at 9 KHz. When you play back this new track it should sound terrible, and the sibilance should stand out as disproportionately loud. Also, nudge this track forward in the timeline so that it plays ahead of the original track by about 5 milliseconds. The goal with this track is to emphasize the sibilance.
Now, set the output of this track to an aux bus, and mute that aux bus. We want this track to play in the background, but never be heard. This is your super sibilant track.
Now on the original vocal track, insert a compressor or a dynamic EQ and side chain it to the super sibilant track. Set your attack as fast as possible, and have a medium release speed, between 40-80 milliseconds..
You can use a regular compressor, which will reduce the entire frequency range evenly. You can still tweak the balance of each frequency using the regular de-essing plugins.
Or, you can use a dynamic EQ. Lesson 55 explains these in regular use. But, the dynamic EQ you use needs to have side chain capabilities. I recommend the Tokyo Dawn Labs Nova, which is available for free from their website.
I prefer using the dynamic EQ, because it provides precise control of which frequencies you are attenuating, and how much. With this, you can use your ears to determine which frequencies need to be reduced.
What makes this strategy so powerful is that there’s no limit to how many sequential compressors or dynamic EQ’s you can use, so you can stack them on top of each other for as much sibilance reduction as you want. Combining this with controlling the balance of the frequencies, even the most stubborn sibilance can be made to sound natural.
3- Using Automation.
The above 2 techniques will work for 90% of the sibilance in a song, but due to massive amount of gain reduction on the sibilance, and variances in the performance, the occasional sibilance won’t sound perfect. Our goal here is for all of them to be perfect, so that’s when you will use this automation, to catch the stragglers which for whatever reason, were outside of the average sibilant sound.
When I need to do this, I’ll have already done the other 2 strategies, so I’ll probably have a couple dynamic EQ’s side chained to the super sibilant track. I’ll automate the threshold on one of these for just the half second sibilance, then return the automation back to normal.
Conclusion
With these strategies, you should be able to eliminate sibilance to a professional level, no matter how stubborn it was.