Understanding Dynamic Processors (Compression)

  • Thread starter Thread starter WeissSound
  • Start date Start date
From what I have gathered this is what I understand plz tell me if I'm wrong.

Threashold = The level the input reach's before compression begins, anything above the threashold will be compessed

Ratio = The amount of compression in ratio, example 3.1 = every 3 db. the input signal reachs above threshold will only rise 1 db. in output to every 3 db. rise in input

Attack = The amout of time it take for compression to begin after threshold is reached

Release = The amount of time it takes for compression to let go.
 
joshs1982 said:
From what I have gathered this is what I understand plz tell me if I'm wrong.

Threashold = The level the input reach's before compression begins, anything above the threashold will be compessed

Ratio = The amount of compression in ratio, example 3.1 = every 3 db. the input signal reachs above threshold will only rise 1 db. in output to every 3 db. rise in input

Attack = The amout of time it take for compression to begin after threshold is reached

Release = The amount of time it takes for compression to let go.
Click to expand...

Close. Attack is the speed of gain reduction. Not the amount of time before the compressor begins. The compressor begins as soon as the threshold is breached. Attack is just how fast it clamps down.

Not to confuse you by adding information - but I'm about to do that. The attack is the amount of time it takes the compressor to do it's initial gain reduction, which is usually 9db (but every compressor is different).

Release is the same as attack but, letting go of gain reduction instead of clamping down.
 
joshs1982 said:
Ratio = The amount of compression in ratio, example 3.1 = every 3 db. the input signal reachs above threshold will only rise 1 db. in output to every 3 db. rise in input
Click to expand...

Allow me to add another detail about the ratio and threshold.

The threshold is not related to the input level, but to the level after the side-chain and the internal AC/DC conversion every compressor needs to create a control signal. This is an important fact. Only very few compressors accurately listen to the input level. You'll often find one or more filters in the side-chain, as well as a stereo-linking structure. Both will change the reference level. Most compressors listen to a smoothed version of the rectified side-chain, sometimes even an RMS average. This means that the threshold can react to something that doesn't really correlate with the input - think about a de-esser or talk-over setup as extreme examples.

The ratio is threshold dependent, so what you wrote about the ratio isn't really true. In most cases, "3:1" doesn't mean that 3dB more input will result in 1dB more output. Even if we had an absolutely clean side-chain and no smoothing, the claim would only be true for a zero attack and zero release setup. But this is a clipper or waveshaper, not what most people call a compressor. As soon you "slow down" the action, the ratio becomes dynamic and will never (!) reach its full value. There's also the concept of "knee", which makes the whole thing even more complicated. For more brain-fvck, keep in mind that some compressors also listen to their output instead of their input (feed-back). :)

BTW, the only "compressors" who really have a clean sidechain and perfectly fullfill what you said about the ratio are limiters. They have to, or they would not be able to prevent overs. This on the other hand also implies that most compressors totally fail at preventing overloads (it was sufficient for tape recording, but not today). Fact is, most don't really reduce the dynamic range from a technical point of view, instead, they even out the long term and mid-term dynamics, while keeping the short-term dynamics more or less untouched.

I know, this is all very technical. But you asked! ;)
 
Last edited:
Nice post! It explains a large part of why certain compressors seem to handle bass heavy signals better, or why some compressors will tend to accentuate certain frequency areas. Those detector circuits can really vary quite a bit. For compressors with an external, or open-architecture side-chain, eq'ing the signal that feeds it can allow for some really precise and effective tone control.
 
adrian8795 said:
That was a lot to take in, lol.
Click to expand...

It is. It's a lot. But it boils down to "shape" and "tone". If you think "shape" and "tone" when you start touching the compressor, it will start to make sense pretty fast. Then the more complex stuff like detector circuits and knee settings can come later.
 
Great info...i needed this....

---------- Post added 09-02-2011 at 07:56 AM ---------- Previous post was 08-30-2011 at 08:14 PM ----------

I would like to see a thread with detailed guidelines for using compression..A more instrument by instrument type of post
 
Hi,
in order to complement what has already been said,
Here is a link to a great video about compression:
Compression Overview - PUREMIX
Fab talks about: La2a versus fader riding, tape compression and talk us trough different uses for a compressor.
Oh! and setting are hidden... so you have no choice but listen to it... :-)
 
moses said:
Allow me to add another detail about the ratio and threshold.

The threshold is not related to the input level, but to the level after the side-chain and the internal AC/DC conversion every compressor needs to create a control signal. This is an important fact. Only very few compressors accurately listen to the input level. You'll often find one or more filters in the side-chain, as well as a stereo-linking structure. Both will change the reference level. Finally, most compressor listen to a smoothed version of the rectified side-chain, sometimes even an RMS average. This means that the threshold can react to something that doesn't really correlate with the input - think about a de-esser setup or creative side-chaining.

The ratio is threshold dependent, so what you wrote about the ratio isn't really true. In most cases, "3:1" doesn't mean that 3dB more input will result in 1dB more output. Even if we had an absolutely clean side-chain and no smoothing, the claim would only be true for a zero attack and zero release setup (but this is a clipper, not what most people call a compressor). As soon you "slow down" the action, the ratio becomes dynamic and will never (!) reach its full value. Oh, and there's also the concept of "knee", which makes the whole thing even more complicated. For more brain-fvck, keep in mind that some compressors also listen to their output instead of their input (feed-back). :)

BTW, the only "compressors" who really have a clean sidechain and perfectly fullfill what you said about the ratio are limiters. They have to, or they would not be able to prevent overs. This on the other hand also implies that most compressors totally fail at preventing overloads (it was sufficient for tape recording, but not today). Fact is, most don't really reduce the dynamic range from a technical point of view, instead, they even out the long term and mid-term dynamics, while keeping the short-term dynamics more or less untouched.

I know, this is all very technical. But you asked! ;)
Click to expand...

Can we have that in Inguish now plz?
 
A compressor doesn't simply read the incoming signal. That's a gross oversimplification. The compressor applies gain reduction to the incoming signal - but it bases that gain reduction on an INTERPRETATION of the incoming signal. Compressors have detector circuits specifically designed to figure out what the incoming signal is really saying and how best to react to it. What feeds the detector circuit is called the "side chain signal". It's essentially a copy of the signal, but it's effected by some degree of filters and voltage step circuitry. In other words, the signal that the compressor "hears" is not really the same as the signal you are feeding it. Reasoning for this varies, but there are three primary reasons: Practicality, giving the end user finer control, and functionality.

The bottom line is that it's not quite correct to say - if you dial in X threshold and Y ratio you will get Z gain reduction. You can set two compressors exactly the same way numerically and get completely different results.

On top of all that, the idea of a "threshold" is not actually a defined line. Electricity doesn't work that way. AND the gain reduction circuitry in different compressors react differently.

This all results in gain reduction that happens with a 1:1 ratio, gain reduction occurring before the threshold is breached, and really never as much gain reduction as one would expect to create with certain setups.

There's a LOT of variables - but one thing always remains the same. The user's intention, and the user's ears. These are very important because the setting on a compressor can yield all sorts of unexpected results.

Mind you - this applies much more to the analog world - but it carries over to the digital world to a certain degree as well.
 
moses said:
Thank you ;)
Click to expand...

Figured your typing fingers could use a break! Good post. I think anyone familiar with the basics of compression who really want to start opening their ears to specific compressors should give your post a few re-reads.
 
This is a great thread - I need to document my own settings more (especially since windows often forgets them;)
 
You painted this perfectly and I would have never thought about compression in this way. Thank you.
 
You must log in or register to reply here.
Back
Top