MAutoDynamicEQ dynamics modes (filtered compensated vs. filtered)

[mesaone]

I've often referred to the SoundBytes in-depth review of MAutoDynamicEQ to answer any questions... But I have one now about dynamics modes - specifically, filtered compensated vs. filtered uncompensated (entire spectrum makes sense; dynamic control of a band takes place based on broadband level of the sidechain rather than frequency-specific level)

Forgive me, I'm just having a hard time wrapping my head around certain aspects here, since my questions require more "under the hood" examination of the plug-in.

The difference between “compensated” and “uncompensated” filter modes is that the compensated mode will boost or cut the frequency band for the sidechain signal to offset the effects of static filtering. For example, suppose you have a high-pass filter that’s (statically) knocking 6 dB off the low end. The sidechain signal would be boosted by 6 dB so that it hits the detector at full volume, unaffected by the action of the HPF.

In that example, I fail to see the utility of compensation. Am I correct in thinking that the node position acts as makeup gain for any dynamics that take place on that node? If, for example, I have dynamics set to -6 dB and my node is positioned to boost by 6 dB, at maximum gain reduction I would have roughly the same level at the center frequency as the input signal, correct?

If I use filtered compensated, my gain reduction would be offset so that maximum reduction would be reached with a quieter input signal.

[MeldaProduction]

Hmmm, I don't really follow too much. It works like this:

Normally the filters are static and work the way you see, nothing so "hidden" about it. Then you make a band dynamic by adjusting the dynamic gain (vertical nodes), the band starts listening to the signal. Now the question is, how! It can follow the whole input signal level (Entire spectrum) or just the part it is actually affecting.

So let's say you create a thin peak filter at say 5kHz to tame some vocal sibilance. You probably don't want it to follow the whole input level as it would be reducing the 5k all the time. You want it to listen to 5k and reduce it when present. In a way like inserting a multiband compressor with a tiny band at 5k. Now here's the problem : imagine you feed it with white noise (it's ideal for explaining it). You create the band at 5k and make it wide (low Q) - it basically follows a bandpass filter at 5k with the same Q, so most of the white noise gets through -> input level is high -> lots of reduction. Then you increase the Q and make the peak filter thin -> the bandpass used by the level detector will be thin as well -> most of the white noise gets filteres out -> input level is low -> not much reduction!

So the "compensated" mode adjusts the input level detector's gain so that it doesn't really matter how wide the filter is. The mentioned scenario is typical and it would be very fiddly to setup the parameters if the input level would be jumping so much, so the compensation makes it much easier. I personally never use "uncompensated" mode.

[MeldaProduction]

Hmmm, I don't really follow too much. It works like this:

Normally the filters are static and work the way you see, nothing so "hidden" about it. Then you make a band dynamic by adjusting the dynamic gain (vertical nodes), the band starts listening to the signal. Now the question is, how! It can follow the whole input signal level (Entire spectrum) or just the part it is actually affecting.

So let's say you create a thin peak filter at say 5kHz to tame some vocal sibilance. You probably don't want it to follow the whole input level as it would be reducing the 5k all the time. You want it to listen to 5k and reduce it when present. In a way like inserting a multiband compressor with a tiny band at 5k. Now here's the problem : imagine you feed it with white noise (it's ideal for explaining it). You create the band at 5k and make it wide (low Q) - it basically follows a bandpass filter at 5k with the same Q, so most of the white noise gets through -> input level is high -> lots of reduction. Then you increase the Q and make the peak filter thin -> the bandpass used by the level detector will be thin as well -> most of the white noise gets filteres out -> input level is low -> not much reduction!

So the "compensated" mode adjusts the input level detector's gain so that it doesn't really matter how wide the filter is. The mentioned scenario is typical and it would be very fiddly to setup the parameters if the input level would be jumping so much, so the compensation makes it much easier. I personally never use "uncompensated" mode.

[mesaone]

Okay, that explains it I think. So for narrow bandwidth where much of the signal is filtered out, Filtered Compensated will boost the sidechain in a way that the dynamics will react in a more natural fashion instead of receiving an artificially low signal.

Thanks Vojtech!

[MeldaProduction]

Yes . My pleasure!