Clean digital compressors

This is pointless.

Consarnit, I'm late to the party and i was looking forward TO THIS READ . . . .

What you are going to find is the cleanest are probably going to be mastering digital mastering hardware ported over to your favorite plugin format BUT, in many cases your going to find that many of these have a distortion component dialed in and that the original developer found this particular distortion profile is what made them popular.

Weiss isn’t even close and is dirtier than what comes with modern daws if the stock plugins are set to behave mildly. Weiss can get more aggressive and not crap out like stock plugs will but it’s still 1990s code at heart with bugs and distortion that give it the driven, punchy distorted sheened sound.

Same with Vulf Compressor. It’s a reverse engineered boss vinyl sim compressor with options to clean it up a bit or get even more aggressive.

The distortion component is present in U-he Presswerk and Goodhertz Faraday Limiter. Turn it way down, turn on HQ, and they’re pretty clean. The powair distortion options rarely work at all and the compressor/limiter portion gets plastic and fills the spectrum with crap on greater gain reductions so it’s probably best as a leveler with a slight peak shaper or limiter atop it but I’ve abused it in the past, probably not for the better.

The best stuff over a wide range of settings were the Tokyo Dawn compressors on insane and Paul Frindle’s Pro Audio DSP DSM. They don’t really crap out or grit up the spectrum and background with partials of partials. The DSM always has minute aliasing but it doesn’t get worse and Molot GE limiter is a soft clipper so having it touch non transients is a recipe for fuzzy disaster

MDWDRC2 can dirty up for the auto release trigger and faster detector that’s set stock as root mean cube. It can get gritty where Kotelnikov is clean and snappy. But the main Detector is very clean when kept as at RMS.

Frindle’s Oxford Dynamics and the ancient Renaissance Compressor (the updated gui sucks. 90s gui for life) are good because they were made to mitigate distortion. So they have proper lookaheads and side chains lowpassed at 2khz to mitigate aliasing. The built in limiters are dirtier, especially R Comp’s L1, which encourages you to straight up run the makeup gain into it as a crappy overshoot limiter 😬 while Oxford Dynamics just uses the faster, harder, dirtier limiter on the higher portions of the signal. The distortion options in both can get messy.

The only analog emulation I had installed that wasn’t scrappy was the Glue but who wouldn’t rather have real hardware instead of an ersatz. Some stuff is cool but it doesn’t work that well on stuff that hasn’t been cooked or that went through a big boy pre and good conversion so all the piled on distorted crap jsut makes it sound worse. Some guys have good tones even DIed and others… 🤯🔫 so they need Vaseline on the camera lens and Omar Sharif.

That’s what the smoothing filter is for 😎
Digital can get the harmonic distortion down to just the third harmonic from the rectifier from filling in the zero crossings and smoothing transitions of threshold. Then just distortion from the smoothing filter being faster than the sine frequency.

Most digital compressors have or gradually become infinite distortion machines that bounce around like crazy. Most Waves emulations and stock plugins are some of the worst ever. 1 khz can produce partials in the mud range and near dc from the random behavior. modulation sidebands those are not.

I agree, I’m not sure low distortion or distortion free is what our ears like. By all means I don’t want needless junk that is a shortcoming of some plug-in coding but I think half the reason people prefer hardware vs a plug-in in a shootout is because they can saturate/distort where often a plug-in can’t.

We have to admit we love distortion!
But we also have to look at things in a very big picture kind of way and realize in the big picture, Dynamics control as we see it’s in fact controlled use of distorted dynamics to begin with ! Any signal manipulation even our choice of mics is as much about controlled use of distortion.

The hardware isn’t saturating. Most of it only saturates or clips well above levels it is typically used at. It just hugs and rides the audio better and this is most noticeable on transients that require faster releases to not pump down the rest of the signal with them. Emulations try to operate on an analog of the analog of the audio and mostly fall flat on their face vs the better analog and digital designs. Analog often (usually) behaves better because the sidechain signal is often (usually) cleaner.

The slight distortion from their amplifiers and transformers can sometime mask some of the distortion from the gain reduction or even serve as a coat of varnish almost but this is often not the case in most modern compressors where the distortion from the gain reduction is usually higher than that from the amplifier.

The analog compressors with noticeable distortion are controversial or of limited utility.

I'm not sure what you mean Dan, tubes, transformers and consoles are all routinely pushed to saturation intentionally as they sound awesome that way.  I push a 176, Sta Level or Bluey there all the time.  Even something as sweet and beautiful as the LH95 EQs have this killer saturation when you pushed into them hot(particularly on a stereo bus).  

In regards to whatever Plug-in doctor is finding, imagine running it on a 1073 vs a modern RND.  The modern one would technically be the much better design but I bet you wouldn't find anyone that thought it sounded better in real world use.  My point is that I don't think low or zero distortion necessarily equals something being better.  The whole plug-in doctor/aliasing thing is a bit weird to me.  If you like the way it sounds I don't know why some readout or frequency plot would convince you something sucks.  Our ears = the ultimate plug-in doctors 😁

thank you for posting this. This was a very great and elucidating post for me.

A poorly functioning plug-in processor will end up guessing. A fast attack digital compressor (or very poor analog one) might not detect the peak you want to compress at all and instead compress afterwards. The peak you wanted to suppress will louder when you raise the makeup gain.

Many functions greatly expand the bandwidth of the signal with harmonics. Analog usually has that bandwidth. CD sample rate digital doesn’t and assumes the signal has already been filtered to the human hearing range. The 90s solution to fit within that seems to have been to bandpass the sidechain to have less high frequency than a telephone. Primitive but effective.

I agree with you and seawell about ears as the ultimate judge just that most of the poorly measuring digital compressors are of a much more limited utility than the cleaner ones. I use some dirty stuff like the omnipressor plugin and fetpressor but if they were cleaned up and modernized to become better behaving on a wider range of their settings, I would be able to use them more often.

:sheepishly raises hand:

🤣🤣🤷🏻‍♀️