Advanced 'auto release' for compressor?

Most of the time you will see two time constants stacked upon each other when a compressor (or limiter) uses 'auto release'.
Would there be a benefit in using this approach:

1 - a timing network with a fast attack an a quick release time, say 30-50 mSec.
2 - a timing network with a slower attack time (30 mS?) and a longer release time, for example 100-150 mS
3 - a timing network with a much slower attack time (100 mS) and a long release time (500 mS)
and maybe even a fourth timing network.

All the outputs of the timing networks are summed, in such a way that the highest voltage drives the control voltage of the attenuator. ( "OR" configuration).
I could imagine that this would produce a very natural sounding compression.
Very short peaks would be compressed, without disturbing the average level, while longer signals with a high level would lower the output for a much longer time.
This would maybe behave more like a 'leveller' or 'gain rider'.
Any thougts about this?

Hello

Seem somehow common, have a look at Studer limiter schematic (like the eurocard version)
SC have 3 timing network in series, for the exact purpose you ask for.
But if you only want 500ms max, two should be enough ?
IIRC the 3 timing nw I'm laking about go up to 5sec !

Best
Zam

The What compressor uses a primitive form of that (2 diodes=2release times)
http://dt.prohosting.com/hacks/what.html
Although Mnats is reasoning on his dedicated site, that optical compressors
exhibit that behaviour anyways (memory effect), it could be interesting to try
with a vca or fet compressor!

L´Andratté said:

it could be interesting to try
with a vca or fet compressor!

Click to expand...

The Studer lim I'm talking about is VCA based
The 169 use this (two RC network) for FET lim with semi auto release



Best
Zam

I have a feeling multiple time constants are the secret to the super-clever compressors like the Compellor, Dominator and the Fatso

The Fatso can remove transients almost entirely, but still have a slowish compressor riding on the bass end of the signal. So there must be a super-fast attack & release limiter in there plus a significantly slower compressor at the same time

How you combine the two must be where the cleverness lives

Nick Froome

A facinating subject...I think Nick is spot on about where the cleverness lies.


There are some very interesting techniques out there...


Safe Sound's P1 white paper  http://www.aybg12.dsl.pipex.com/downloads/p1whitepaper.pdf


Three stage Release is discussed in


BBC R&D Monograph 70 http://www.bbc.co.uk/rd/publications/bbc_monograph_70
http://www.bbceng.info/ti/eqpt/AM6_14.pdf
http://www.bbceng.info/ti/eqpt/AM6_7.pdf


The Compellor has been a favorite of mine for years.... I wouldn't mind betting that the Aphex VCA's control curve contributes to the smoothness of the device.... Biamps installation equivalent device went to some lengths to echo this iirc.. 


Another favourite is the Aphex 2020 with its 'sticky leveller' and FDL


Other 'old' techniques are the delayed release used in the NTP 179-160


Good luck with the voyage of discovery


tc

That is some very interesting material, thanks!

I read the Safesound white paper - interesting. Clearly some companies do original research to come up with ways to mitigate the sonic impact of their compressors

After which they hand them over to engineers (like me) who abuse them to get dramatic results

I have a lot of respect for Ray Dolby.  He created time constants for the companders in Dolby A and SR which seem to work pretty much imperceptibly. Thinking about it, the four frequency bands of Dolby A might have been chosen as much to minimise the audible negative effects of the NR process as to maximise the perceived degree of noise reduction

Nick Froome

Sounds a bit like the GML8900 with its three detectors, interrelated attack and release times, and hysteresis?

https://www.massenburg.com/wp-content/uploads/2012/10/8900OperationSummary.pdf

There's also a video on Youtube showing Mr Massenburg explaining the unit.
Just saw on their website that the unit is 'discontinued' 

pvision said:

I have a lot of respect for Ray Dolby.  He created time constants for the companders in Dolby A and SR which seem to work pretty much imperceptibly. Thinking about it, the four frequency bands of Dolby A might have been chosen as much to minimise the audible negative effects of the NR process as to maximise the perceived degree of noise reduction

Nick Froome

Click to expand...

Even more so, I would think. The degree of NR required was not huge, much more important was it's lack of footprint.