Frequency Dependent Compression

[heavymetalmixer]

I watched a Dan Worral's video about "Emphasis and De-emphasis EQ", and got interested in that technique. I've also seen some people call it "Frequency Dependent Compression" when a compressor is used in the middle (do not confuse with multiband compression nor dynamic EQing).




Now, I have some questions:
1) Is it really this or a different concept?

2) How do you use it and why?

3) Any details, quirks or gimmicks you know about it.

4) What plugins do you use for it?

[illegal4Hunna]

This is basically sidechain compressing. It's actually really useful for a lot of things, especially for vocals or midrange instruments when you wanna keep the original tone in place but stop it from getting too unruly. I tend to use it most on nasally or pokey vocals, where EQing out the harsh frequency would dull the sound too much. Most of the time I'm using it with something clean like FirComp or Fabfilter Pro-C 2.

Also pairs well with saturation plugins. Sometimes you need more low mid weight and you can use this to do that .

[cron]

This is basically sidechain compressing.

Indeed. We usually think of sidechain compression as using one signal to duck a completely unrelated second signal, but instead sidechaining with a subtly manipulated (e.g. EQed) version of the signal being compressed can be very useful.

The obvious example with compression would be if your compressor is over reacting to bass signals. A lot of compressors have a low cut built into the detector to help you deal with this, and it's equivalent to sidechaining with a copy of the signal with the bass rolled off.

The other obvious application would be to make a de-esser. You'd take a copy of the signal being compressed, filter off everything but the high end S frequencies, and feed that into the sidechain. The result is a compressor that only ducks on S sounds and leaves the rest of your vocal untouched.

[lobanov]

Every compressor (or in most cases) has two audio streams. The first one is main (is being compressed), the second one is a sidechain (for detection of peaks). Usually they are identical but if they are different we name it 'sidechain compression' explicitly. This means that every compressor has the sidechain line for measuring. And we can cut or even boost frequencies on this side channel. This is a standard way to design compressors.

I think that a properly designed frequency dependent compressor should work this way. So we don't need the second stage (deemphasis or reemphasis). Sidechain is needed to fire attenuator on the main audio stream only.

But this is my assumption.

And nothing prevents us from doing this as with this video.

P. S. The music in this video so catchy that I can't listen to Dan' narration

[Rockatansky]

It's a pretty common trick to do this using tilt EQs (essentially two wide shelving filters at same cut-off frequency, one boosting and one cutting by the same amount) around distortion, saturation, clipping, et al. Distortion essentially generates new high(er) frequency content, so if a lot of low end signal is distorted, the newly generated harmonics would litter the low mids/mids, and obviously the high frequencies as well. With a tilt EQ before distortion, the lows can be dampened and the highs can be emphasized, which makes the distorted signal cleaner in the mids and highs. Then after the distortion another tilt EQ can invert the work of the previous one, and bring back the lows, lower the highs. This will not only make the lows litter the midrange less, but it will also lower all the new high frequency fizz and buzz in relation to the rest of the lower frequency spectrum, which will usually sound softer or smoother than just blasting the entire spectrum through a distortion processor. Especially great for guitar DIs, because it focuses the distortion on the high frequency transients (pick attack), keeps the midrange crisp (snappiness), and doesn't turn the lows (signal body) into a garbling mess.

[jens]

The king of sidechain-equalization:

https://www.bozdigitallabs.com/product/ ... ompressor/

Not right now, but it's often on sale and can then be had for a lot less. And it's a great compressor overall.

[anissbenthami]

Frequency dependent compression isn't emphasis/deemphasis nor sidechain, it's compression that reacts differently to different frequencies of the signal going through, a kick drum would be compressed differently from a snare drum even under the same settings (attack/release, ratio, curves...)

[imrae]

Frequency dependent compression isn't emphasis/deemphasis nor sidechain, it's compression that reacts differently to different frequencies of the signal going through

Emphasis/de-emphasis and sidechain EQ are more-or-less-equivalent ways of achieving frequency-dependent compression. Every compressor reacts to level, and those processes modify the relationship between frequency and level - and, therefore, the relationship between frequency and gain-reduction.

It's worth mentioning there is another blend of frequency/compression available than sidechain manipulation and split-band approaches. Kotelnikov GE has a "frequency-dependent ratio" feature which is the opposite to sidechain EQ: the gain reduction is calculated as normal but then the "delta" signal is equalised so that not all frequencies are reduced equally.