Dynamics Processing - Level Detector & Gain Processor: Why 2 Attacks & 2 Releases?

Report · Jan 04, 2019

Context - I have been an audio engineer since the mid-90s. I learned on Analog gear. I got into DAW recording in 1997. I have extensive knowledge of how compressors and noise gates and expanders work, so I am not looking for an explanation of compressors. However, I am having a difficult time understanding the user interface of the Dynamics Processing plug-in.

I feel like I grasp the main graph on the "Dynamics Tab," but I'm having a hard time understanding the Settings tab.

Under the settings tab you have the Level Detector section and below it, the Gain Processor section. Level Detector has an Input Gain slider (this makes sense). The Gain Processor has an Output Gain slider (this makes sense) as well, and both have attack and release settings.

MAIN QUESTION: I understand what attack and release settings are for on any other compressor/gate/expander, but why are there two sets of these settings under this tab? I notice there is no "hold" setting. Can someone help me wrap my old brain around this weird looking compressor?

I've got to admit, I'm not a big fan of this plug-in because I feel it over-complicates the traditional and fairly sublime interface of regular compressors, but I've got an open mind. I just want to understand how it works because right now, I'm having a hard time getting it to do what I want.

I know the new version of Audition has new processors, but that will not help me. My job takes a couple of years to adopt new updates sadly. I want to understand this plug-in specifically.

Report · Jan 04, 2019

Okay - true confessions time; we've always had some difficulty with this. It's a hangover plugin from CEP (Audition's predecessor from Syntrillium) and back then we got confused by it, even with a bit of help from the guy who designed it! What I can tell you though is how it's supposed to work, why it doesn't appear to have a threshold control (it doesn't need one - it's graphical), and perhaps explain why this makes it so confusing:

On the Settings tab, you have a Level Detector, and a Gain Processor. The Level Detector is the part of the plugin that analyses the incoming signal, and lets you boost it if it's weak. This is a Bad Idea - you are much better off leaving this at 0dB and normalizing your signal before applying it - you'll have a much better idea of where you are then. The Attack and Release times here only relate to how your incoming signal is analysed. The Gain Processor lets you determine what the attack and release characteristics of the processed signal will be, and its output gain control effectively controls the amount of 'influence' the driving signal has on the output. Yes, all these attack and release controls appear to conflict with each other, and the output level tab is far from intuitive. As far as the other controls on that tab are concerned, the lookahead time should be obvious (helps you catch transients better) and the Band Limiting and Noise Gating likewise.

To get some sort of idea of how much these Attack/Release settings don't really matter so much, leave the tab open and flick through the presets - and you'll notice that nothing really changes that dramatically unless it's for something excessive. Basically, unless you have good reason to alter them, leave the settings pretty much where they end up on average - attack times low, release times around 2-300ms.

What will change though between the presets is the graphical display, and that's how you set the basic system up. What's really clever about it is that you aren't simply limited to a single threshold - you can set up multiple break points and have all sorts of curves, depending entirely on what you want to achieve. Within the limitations of it being a single-channel processor, pretty much anything is on the cards if you can set it up in six segments - which is why it's still here, and still popular. Even more so now that it has a gain reduction meter!

You will notice that you can also spline the curves, and that the amounts of compression and the appropriate threshold levels are displayed for each segment. You can also set the makeup gain from this panel. Once you've got something you like, save it as a preset, or you may never find it again!

Bottom line - perseverance pays off with this thing!

Report · Jan 04, 2019

DUUUUDE! First, thanks so much for such a detailed explanation. I have been trying to find and answer to this for a long time. I gave up on the thing the first few times I tried it a couple years ago mainly because I could not understand what the point of the complicated interface was for what was in the end, a somewhat sonically lackluster - tho usable - compressor. But your explanation makes it as crystal clear as I think it can be from a technical standpoint.

Playing with the actual controls on different types of audio material, it's hard to understand why two sets of controls are useful (instead of just confusing) for the normal sorts of things people use dynamics for. In other words, why not just hide one set (probably the Level Detector) of controls and just have that setting as something usable and constant built into the code, and just leave the second set for tweaking? Why allow the user this extra control? I can tell they interact with each other in some way. Once I start dialing in some extreme settings on both, it becomes obvious both are having an audible effect on the audio. I'd love to pick the designer's brain about this. Can you think of a practical situation where this could be helpful? I will definitely be playing around with it tomorrow.

For work though, my instinct is just to just set and then ignore the attack and release controls to something standard-ish (0ms attack, 100ms release) on the Level Detector and just tweak the Gain Processorwhich, like you said don't really matter a ton if they're being used within practical limits and not trying to do something extreme for compression. They matter more for noise gates in my experience. The effect of the gating can have much more obvious results (sometimes very negative) when set poorly, especially for human voices with all of the lingering word fall-offs like on S's and such.

I have more respect for this plugin now, although I still probably will never use it because I have some better sounding ones anyway, but a guy I work with who is new to compression likes it and I also like the graphical representation of what's happening. I just wanted to make sure I wasn't too quick to write it off due to my ignorance about its design.

Thanks again for you answer. It was extremely helpful. 😃

Report · Jan 04, 2019

One other question - Is the "Makeup Gain" on the Dynamics tab doing the same thing as the "Output Gain" in the "Gain Processing" section of the Settings tab? Again, it seems like a redundant control, but still separate - as in tweaking one has no effect on the other from a visual standpoint on the interface, but they are clearly both affecting the audio independently of each other. Independent, but intertwined as in, if you boost 10db on the makeup and boost 10db on the Output Gain, you're gonna end up with 20db of boost.

Again, on the face of it this just seems redundant even though they both must be located at different points in the signal chain as the audio crunches its ones and zeros through the processor. My guess would be the Makeup Gain is first in the chain and Output Gain second, but is this a known thing? Is there a practical purpose for how this could be useful in some situation?

Report · Jan 04, 2019

!QAZ2wsx wrote
One other question - Is the "Makeup Gain" on the Dynamics tab doing the same thing as the "Output Gain" in the "Gain Processing" section of the Settings tab? Again, it seems like a redundant control, but still separate - as in tweaking one has no effect on the other from a visual standpoint on the interface, but they are clearly both affecting the audio independently of each other. Independent, but intertwined as in, if you boost 10db on the makeup and boost 10db on the Output Gain, you're gonna end up with 20db of boost.
Again, on the face of it this just seems redundant even though they both must be located at different points in the signal chain as the audio crunches its ones and zeros through the processor. My guess would be the Makeup Gain is first in the chain and Output Gain second, but is this a known thing? Is there a practical purpose for how this could be useful in some situation?

Well, the makeup gain should come last; it's compensating for any reduction in the max signal. The output gain is claimed in the blurb to do exactly the same thing, but altering it on the graphical tab doesn't alter the slider setting, so I'm not so convinced that this is correct. In the blurb it says:

Output Gain Applies gain to the output signal after all dynamics processing.

Make-Up Gain Boosts the processed signal.

Good luck with trying to work out what the difference is... What I suspect that it might be is that the Output Gain somehow makes the gain cell's actions more 'aggressive', but even that doesn't quite seem right, as that's what the graphical controls are for. All attempts to get to the bottom of this have failed in the past, and the person who wrote it is no longer around in Adobe. But as I said, that doesn't make a lot of difference because when we asked David about this years ago (about 15 of them), we didn't get a sensible answer even then, so I reckon there's even less chance now!