The official website showcasing the music of Yellowgold, produced by Jason Howell.

Filtering by Category: Signal Processing

Better DeEssing

Don't just de-ess your entire vocal track. Automate and use it only when its needed. The utility of a DeEsser is normally rather specific, as it is a pinpoint way to compress a particular frequency range. So, as its name implies, its good at removing the harshness associated with the sound of the letter S, among other things. The thought being that you apply the DeEsser to a track with problem S's, find the frequency, then dial in how much of that problem frequency you wish to reduce to an appropriate level.

I've found that it can sometimes be hard to know when you actually need to use a DeEsser. First, your monitoring situation might not be ideal. Good monitors expose these kinds of issues, and bad monitors hide this and much much more. Also, this is particularly hard when you are entrenched in a long session and your ears are already fully adjusted to the sound of your track. Your ears might simply be used to what its hearing. So I tend to act on the need to de-ess when I'm approaching a track for the first time during a session.

The problem with a DeEsser is that it affects the whole sound when applied in a blanket sort of way. So, say you find the frequency to adjust, and simply apply the plug-in to the entire vocal track. When there are NO S's to knock down, it's still affecting the sound of the rest of the vocals. It dulls the sound, likely, and removes some of the clarity. As a result, I usually apply DeEssing in a manual way as opposed to simply dropping in the plug-in and moving on. And yes, this takes a while, but in my mind its worth the effort. So let's take a look at the process, step by step. In this example, I'm using the Waves Renaissance DeEsser, but any will do if you follow the general idea:

  • Place RDeEsser on vocal Comp track
  • Enable automation for the Range control
  • Load Male DeEss Narrow preset as a starting point
  • Turn on Side Chain and solo the vocals you are treating (this allows you to ONLY hear the audio that will likely be affected by the DeEssing)
  • Sweep to find the harshest frequency in your most problem S area
  • Once found, switch back to the Audio setting
  • Adjust the Threshold so the offending S's move the reduction line in the graph. (We will be automating this anyways, so don't worry if it moves when none-S's are audible too)
  • As a quick experiment, pull Range all the way down to hear the extreme. S's turn into lisps when you do this, likely. Sometimes when dealing with effects like this, it's good to test what the extreme setting gives you. Then you have a basis to compare against when trying to set it properly.
  • Now, reset range to 0 and pull it down slowly only until the harshness goes away. You shouldn't need to go very far, hopefully.

Enable automation for the Range control

Testing with extreme range

Write the Range number down!

Now, THIS is where the range needs to be in order to tame those S's. WRITE THIS NUMBER DOWN.

We aren't done yet. The last thing you want to do at this point is just leave the plugin on with these settings. The quality of the rest of your vocals (i.e. the parts that aren't S's i.e. EVERYTHING else) will be affected by this. You can preview the rest of your vocal track and toggle the Bypass on the plugin to see what I mean.

Here's where we automate the range:

  • Set range back to 0 to start off with.
  • Play your vocal track until the very first offending S, and park your cursor slightly before that point.
  • Turn your track view to show the automation track for the RDeEsser Range function. (It should show that its currently set to 0 across the entire track)
  • Draw in the automation AROUND that SS as shown in the screenshot, making sure that at its most dramatic point, the automation hits the Range you wrote down above.
  • Now you've automated the DeEsser to only knock the S down in the range that you specified and at the time that you specified. Afterward, it resets to a range of 0 effectively bypassing the processing.

Drawn automation for the first S

Now, we get the payoff for all of that setup:

  • Highlight that automation you drew around that one S and copy it.
  • And now, play through your track, and anytime you encounter a bad S, paste that automation in slightly before the offending S. If it's not perfect, adjust as needed. The point being that you've created the automation block to treat your S's, so pasting that automation block wherever you encounter a bad S should get you most of the way there to treating them pretty efficiently.
  • And as always, different S's might require different levels of range (or even different frequencies!) So how detail oriented you want to be with this process is up to you and your patience.

Automation pasted onto other problem areas

What you've done here is utilize all of the benefits of De-essing without blurring the crap out of your entire vocal line in the process. You are automating the effect INTO the track ONLY as needed.

Now for some examples from The Last Thing I Ever Do, one of the tracks off my upcoming album. Here are the various steps I got to as I walked through the process listed above:

Hope that all makes sense!

Reader comment: Noise gate versus manual editing

In response to my previous post "Eliminating noise in your project", Ronnie Marler asked this question:

"How is this different/better compared to the use of a limiter / noise gate plugin?"

This is a terrific question, and I suppose the answer comes down to personal preference. Noise gates allow me to set a threshold below which no noise will pass through. So, for example, if I set the threshold slightly above the perceived noise floor, the Noise Gate will not pass any audio through the signal chain, thereby eliminating the noise automatically. The second I begin to sing or speak, that threshold is passed and the gate opens to allow the audio to pass through. Once i stop talking, the threshold is passed on the way down to the noise floor, and again, the gate closes letting no noise through.

Here's my take on it. Noise gates are fantastic tools, and particularly in live environments, they are essential in many ways. In a studio environment, however, you never really know exactly what you are getting in the mix.

For instance, if I set a noise gate on my vocal line and set it up as close to perfect as possible, there is always the chance that something gets through that shouldn't. Or, even worse, something doesn't get through that should.

Let's use the breathe in a vocal phrase as just one example to illustrate what I mean. When dealing with vocals, for example, sometimes you want that gasp of breathe leading into the chorus to be there. It makes the vocalist sound more human. Sometimes those breathes in between phrases gives it a more human feel and brings the listeners closer to the vocalist.

Think about what a single gasp of breathe sounds like. It's not a sudden noise that goes from silence immediately into full volume breathe. A gasp of air actually ramps up gradually from silence (before the gasp begins) up to the full volume breathe. Looking at a waveform of this, you see the peak of the ramp where the cursor is dropped:

Breathe leading into a vocal phrase

Breathe leading into a vocal phrase

If I use a noise gate, I am saying "this is the point to let audio through" and in the case of this single breathe, that threshold by design is passed sometime after the breathe has already started ramping up, in essence, resulting in only PART of the breathe getting through and not the full breathe. This might be what you want, and with the added controls of attack, release, and range, you can likely tweak this so no one is the wiser. But my instinct is never to reach for a noise gate plug-in in this case simply because I'm fine doing the work manually, thereby knowing exactly how each part is treated.

I know, plenty of people use them all the time to great effect, and I'm by no means saying that its the wrong way to do this. (who am I to tell ANYBODY that they are doing any of this wrong? It's all about experimentation and finding what works best for you.) In fact, I'm pretty certain some of you are yelling at your screen, telling me I'm completely wrong here. But as I said, it all comes down to personal preference. I'd much rather take the extra time and treat each phrase manually through my own editing than set up a number of noise gate plugins to hopefully be smart enough to do this without any error.

Not to mention, I like to try to keep my plug-in count down as much as possible. If I make these edits manually, I save my processor for more intensive things.

Ultimately, I'm simply not a big fan of processing things with automatic settings if it can be done manually. YES, it takes more time. YES, it can be incredibly repetitive and boring. But for me, it's time well spent because I have granular control on a point by point basis of what the track sounds like at any given moment.

In fact, this leads perfectly into my next post dealing with De-essing vocals. De-essing is a process that is often used in an automatic way, but I choose to  apply it manually as needed. I'll show you how I approach this next. Stay tuned!