Stereo Tool

AFAIK most AGCs (the AGCs that have crossed my way and as far as I remember) work this way:
The level of the input signal is detected and if it is below a given value the AGC applies gain untill the desired level is reached.
If the input level is above the given value the AGC applies negative gain (reduction).

And I remember when I first tried to understand and use Stereo Tool's AGC I had the problem, that it seemed as if the AGC was dead (not working at all).
My problem was:
I have tried to use the AGC only (without any Pre Amp !) by pressing the reset button and then enabled the AGC only and played especially tracks with extreme low volume.
And what happened (happens) ?
Nothing.

The Stereo Tool AGC can reduce the input level (fed by Pre Amp gain) and later raise the level again if needed.
But it cannot raise a low input level (and never above the level of the output from Pre Amp) !
It only reacts of the Pre Amp gain.
So there is always this extra step of applying gain with Pre Amp independant of the input level of the source.

Massive gain is applied even for very loud signals and in next step the AGC reduces it again (this happens 90 percent of the time).
Isn't this an unnecessary extra stage of compression that better should be avoided and maybe the root of all evil ?

So instead of this Pre Amp AGC combo only use another type of AGC ?
An AGC with built-in adjustable target level and gain rise as first stage for the input (now we have only a target output level) ?

If this is all naive nonsense (from a person that do not like AGCs ), then forget about this.

You may be onto something here. Let me think aloud a bit...


Right now, I lower the output level if there's too loud input. So, in a track with constant RMS level, I lower it to the appropriate level; in a track with wildly varying RMS level I lower it at spikes - with, if the Up Speed is set too low, the risk that the overall value gets too low. (And setting up speed higher may cause pumping).


Now the opposite: If the level is too low I increase it, but I never lower it. (The big question here is: is this different).

For a constant RMS track I again raise the level upto the requested level -> OK.

For a wildly varying RMS track I increase the volume during valleys. But on a loud peak I lower it again -> hm, similar to what I did before. This is not the solution...


Using both negative and positive gain -> This also sounds reasonable, but in the end it would behave the same.


The issue here is that loud peaks need to be lowered very very rapidly. Gaps must NOT be increased rapidly. There is an imbalance between these two...


Some other AGCs lower the volume far less rapidly, (and you can also configure ST that way), but that causes other annoying effects (on loud bursts the volume stays loud for too long).


What could work - and this is also what Bojcha asked, and what I had a few months ago - is having 2 AGC's, one that responds very slowly (both up and down), and then a 2nd one that responds much faster (which causes some pumping), but normally doesn't need to do anything unless there's a sudden volume rise.


...still thinking.....

Current AGC is good!!

What is needed is 2nd AGC with faster down and avarage up-speed, or Sidechaned compressor which actually controls this AGC always as needed.
Nice example is "ClawFinger - Do What I Say". In my "Addiction" preset it's only Half job done here. 2nd AGC reacts good, but sound is already too late down in 1st AGC and multiband and for a 1/4sec sound is louder. (at start of louder part)

Final thing is Sidechained EQ in AGC and in 2nd AGC! 2nd AGC need this function more! Why?
Let's say you have some track with bass beat louder then anything else. It will constantly "activate" 2nd AGC and push rest of sound down too.

AGC should react more on freqs from ~300 to 3kHz. so, on that freqs AGC will react more.
I already posted pic from BBP, where AGC at 40Hz reacts less by almost 15dB then at, let's say, 1000Hz.

Parametric EQ will be even better.