DIY specialty plugins: the de-esser. Learn to make one!
Tired of your regular old plugins? They not offering you enough versatility or possibilities? Don’t like how they sound?
Listen, there’s a lot of plugins around. An incredible amount. So much, in fact, that I wouldn’t dare putting a number on it. They’re all pretty fun and most of them offer really good sound quality.
But if you’re like me, you know that getting every plugin you ever wanted can get pretty expensive. Especially if you want to get a specific plugin for a very specific task.
So what if I told you that there’s an alternative?
What if I told you that within your DAW resides the power to combine basic plugin functionality in order to build much more powerful and fully functional specialty plugins?
That’s right. It’s possible to build your own plugins within your DAW.That’s because a lot of things like multiband compressors, de-essers, multiband saturation etc. are in fact a rather clever combination of using different basic plugins such as EQs and compressors together in fun and exciting ways!
I’ve got twitter user j operator to thank for his suggestion to write an article on a DIY (software) de-esser. I thought it would indeed be an excellent idea to teach all of you how to build your own de-esser with the tools you already have!
Now while you might already have a de-esser in your arsenal, it’s always nice to know we can build one ourselves just as easily. In fact, we can build one that is tailored specifically to the needs of the track at hand!
Let’s start by looking at what a de-esser actually is, before we dive into how to build one exactly.
What is a de-esser?
A de-esser is one of those tools that can be a lifesaver in your mix. Especially when you’re mixing vocals, you’ll probably have come across a de-esser like tool to reduce harsh sibilance (“ess” or “shhh” sounds).
I could do a lengthy post on what exactly a de-esser IS, but there’s a pretty good description of a de-esser on wikipedia. You can read up on it there, but in essence, a de-esser literally does what the name implies: it lowers the prominence of all the harsh sibilance sounds in a recording.
How to use it? It’s simple, really. You just tell the de-esser ‘this is what the “ess” in my recording sounds like, please make it less harsh without affecting the rest of my recording!’… and then it goes to work.
How does a de-esser work exactly?
Put (very) simply, a de-esser needs to be able to do two things:
- Detect harsh “ess” sounds
- Lower harshness (or prominence/volume) of that specific “ess” sound, when it occurs.
The mechanism behind these two steps is in essence a combination of EQing and compression:
- Select frequencies where this specific “ess” sound is most prominent: this is your “ess” detector
- Lower the volume of this “ess” once it’s detected, using a compressor.
How to make one ourselves?
Now that we know what exactly a de-esser is and how it works, it shouldn’t be too difficult to make your own, right? RIGHT!
Let’s look at how to do it in two different DAWs. I’ll start off with Reason, just because the modular workflow of Reason makes it very easy to understand the basic signal flow and concepts. Once you understand how it can be done within Reason, it’s easy to transfer the concept to another DAW. Just for completeness’ sake, I’ll also talk you through setting one up in Logic Pro X.
DIY de-esser in Reason
Suppose we have a vocal, which contains a harsh “ess” sound. In Reason, this signal would flow through a mix channel, which means that we would be able to add in the effects directly.
However, for the sake of this example, we’ll create the de-esser within a combinator. This way, if you follow along you can save the patch, customize it and re-use it later.
In Reason, start off by creating a combinator.
In the combinator, you’ll need to create a spider audio merger/splitter. Flip over the rack and route the “to devices” output of the combinator to the audio splitter input of the spider.
Next, insert two instances of MCLASS EQ and one MCLASS compressor. Remember to hold down shift so no connections are made automatically. We’ll be creating the connections manually.
Now, we’re going to need a method of detecting the “ess” and feeding it into a compressor. That’s fairly simple. Start by routing one output of the spider splitter to the first equaliser’s input. The output of this first equaliser will then go to the input of the compressor.
Let’s name this first equaliser “analyser”. You’ll use this to analyse the audio and detect the “ess”. You can do this by enabling both of the parametric EQs and setting them to exactly the same value.
For this example, put the freq of both at 4.000hz, gain to full and Q at half. You’ll notice a steep rise in the graph at around 4.000hz.
Now, we don’t want this analyser to trigger by any low-end information. So enable the lo shelf, put freq at 600hz (full) and drop down the gain entirely.
Getting all this? If it’s confusing, here’s what the front of your rack should look like right now:
OK, let’s move on. Remember how a de-essor works in principle? It detects certain “ess” frequencies and lowers the volume when an “ess” is detected. Our first step is completed already: we’ve detected the “ess” sound and sent it to the compressor.
Once it’s detected, we need to somehow lower the volume of that “ess” specifically. This is where it get’s sort of tricky, but bear with me. On the back of the compressor, you’ll want to route the “gain reduction CV out” to the CV input of rotary 1. Why? That will become apparent in just a moment.
While you’re on the back of the Reason rack, connect the second output of the spider splitter to the input of the second equaliser you created earlier. Now name this equaliser “filter”, just so you don’t get lost later. Route the audio output of the “filter” to the “from devices” input of the combinator
Here’s what all your routings should look like right now:
Flip the rack back to the front view. Now on the combinator, click “show programmer”. Here’s where the magic will happen. Select the “filter” eq and on the right assign rotary one to effect > parametric 1: gain. Important: set the min: to 63 and the max to -64. If you don’t, you’ll be exacerbating the “ess” instead of reducing it.
Finally, to use the de-esser we’ve just created, connect the combinator input to the mix channel of the audio you want to de-ess. Create a new mix channel and connect the output of the combinator to the input of that. (Or save the patch and insert it into the audio mix channel as an effect. Same result.)
OK, let’s take a step back and analyse what will be happening here.
The signal from the combinator gets split into two parts. One part goes to the “analyser”, the second part goes into the “filter”. The “analyser” detects an “ess” sound and feeds it into the compressor.
Every time an “ess” is detected, the compressor will trigger, sending a gain reduction CV signal to rotary 1.
Because rotary 1 is connected to the parametric 1: gain function of the “filter” equaliser, it will lower the gain of this parametric EQ proportionate to the amount of gain reduction that the compressor tells it.
So if you dial in the parametric 1: frequency to the specific “ess” sound you want to get rid of, it will automatically be reduced in volume.
And …. That’s it really! You’ve succesfully created your own de-esser. Nice!
The only thing that’s left is for us to dial in the amount of volume reduction we want for the “ess” sounds, by configuring the compressor. Now, how a compressor works is a whole other subject entirely and I’ll be talking about it in the Start To Mix series in detail. To be honest, you shouldn’t be messing around with DIY de-essers if you don’t understand how compressors work.
For effective de-essing, it’s very important to specify the attack: how fast will it lower the volume? And the release: how long will the volume be lowered for? For de-essing purposes, you’ll want to set the attack and release times so that only the harsh “ess” sounds are affected and nothing else. Usually, that means fairly fast attack and release times, but it really depends on the source signal.
Now, the cool thing about this is that you get a tremendous amount of control. You can really dial in the specific sounds you want to temper and if you want to have even more control you’re always free to process the sidechain signal futher before feeding it into the compressor. You can go crazy with it. So much possibilities!
A lot of work, but worth it!
To make detecting easier, you can configure the combinator rotaries to the param 1 and 2 frequencies of the “analyser” EQ. To easily control the amount of de-easing, map one of the rotaries to the threshold of the compressor. Finally, don’t forget to set a ratio that is suited to your audio.
Now let’s look at how we can accomplish something similar in Logic Pro X.
DIY de-esser in Logic Pro X
We’ve covered how to make a de-esser yourself in Reason. In essence, the same principles apply when transferring these techniques to another DAW.
However, I’d like to go over one way of creating a de-esser within Logic. This technique makes use of a duplicate audio channel to detect the sibilance and a sidechain compressor to lower the harshness. This is generally how you would do it in just about any DAW. Explaining it with Logic will get you on your way to doing it in your own DAW. The process is fairly straightforward so I’ll be going over it more quickly. The concepts should be clear to you by now.
- Make a copy of the audio you want to de-ess.
- Name the newly created audio track “s sound” and set the output to “no output”. That last bit is important, because if you don’t eliminate the output, you’ll be exacerbating the “ess” rather than removing it.
- On the “s sound” track, create an EQ. This can be any EQ that you like, but the standard channel EQ works just fine.
- Isolate the harsh “ess” that you want to get rid of or lower in volume:
- Going back to the original audio that you want to de-ess, insert a compressor. This can again be any compressor you like to use – provided it has a side-chain input! I’ll be using the stock compressor so everyone can follow along nicely with this example.
- In the side-chain input of your compressor, select the “s-sound” audio track you created earlier. In my case, this is “audio 2”.
- Dial in your preferred compression settings. As we talked through in the Reason example, we’ll want a short attack and release time and a fairly high ratio. Lower the threshold to your liking, so all the “ess” sounds get lowered in volume. Don’t forget to set the detection mode to “peak” rather than RMS. We only want to remove the “ess”, nothing else.
As explained when we created the de-esser in Reason, the side-chain compressor will only act when it gets an input from the “s sound” bus that we created. Since this “s-sound” bus isolates the “ess” and feeds in into the compressor, it’ll effectively lower the volume of the source signals’ harsh “ess” sounds every time it occurs.
There are a lot of different ways to go about de-essing. In this article, we’ve only scratched the surface of what’s possible. De-essing is a subject where there’s a lot to learn and to watch out for. For example, you really want to watch out you don’t overdo it. The result of too much de-essing would be a lisp effect, which you really want to avoid if you don’t want your singer to come shout at you the day after and break a lot of things in your studio.
Anyway, I want to finish this off by pointing you guys to some further reading. There are a lot of interesting articles on de-essing on the web, but here’s two that caught my attention:
The SoS article is quite a bit more advanced compared to what we talked through here so I’d suggest reading that one when you’ve got a full understanding of how to use an EQ and a compressor together to de-ess. Modern Mixing takes de-essing techniques further, by talking you through how to use it to get rid of some problems in the lower end. Very interesting indeed!
So, let me know in the comments: what are your favorite de-essing techniques?
Thanks again for letting me teach you how to improve your sound!
Kevin – The Soundcoach