Page 1 of 3

Summing Left & Right but lowering Center information

Posted: Wed Jan 15, 2014 4:09 am
by vintagedesign
Is there a simple way to sum L & R and still be able to lower the centre info?
Like to try this in a compressor design with a single sidechain where L&R are summed before the threshold detector.
Mono-ing the signal after a M/S wide control will not work as it's not a mono compatible process

Re: Summing Left & Right but lowering Center information

Posted: Wed Jan 15, 2014 7:57 am
by JR.
If you sum L+R with L (or R) swapped around to opposite polarity, the center infor will be completely gone.

It seems an adjustable phase shift network that gives something less than 180' would only part cancel the center information .

I don't know how to make a phase shift network without a frequency dependency but you probably want to dip the mono preferentially at low-mid frequency so that might work.

Good Luck..

JR

Re: Summing Left & Right but lowering Center information

Posted: Wed Jan 15, 2014 9:11 am
by vintagedesign
Thanx JR.
Will try that before I use dual detectors and just sum the control voltage with diodes instead.

Re: Summing Left & Right but lowering Center information

Posted: Wed Jan 15, 2014 9:14 am
by JR.
You could do a L+R and a L-R, then pan between them or selectively use a fraction of each to vary the L+R contribution.

JR

Re: Summing Left & Right but lowering Center information

Posted: Wed Jan 15, 2014 9:25 am
by vintagedesign
The side info will then also change. What I'm after is to have a -6dB for the centre info.
0dBu on left or right will hit the detector at 0dB but if it's precent on both channels it will be +6dB when summed.
Am I wrong?

Re: Summing Left & Right but lowering Center information

Posted: Wed Jan 15, 2014 9:51 am
by JR.
If you perform L-R but imbalance the gain of the two, say L-(R/2), this will knock down the M -6dB while pure R will be down too.

JR

Re: Summing Left & Right but lowering Center information

Posted: Wed Jan 15, 2014 4:48 pm
by mediatechnology
Mono-ing the signal after a M/S wide control will not work as it's not a mono compatible process
It's been a long day so I need to think about that. Are you sure?

I'm reminded of this thread: viewtopic.php?f=6&t=110&p=890

What's discussed there is not what you're trying to do, but in the above example I "created some stereo" and it does fold completely back down to mono. Granted it's a different process.

Assuming that the side/stereo elements in the mix are mono compatible themselves I'm not sure why taking out some mid and then summing the result wouldn't be as mono compatible as the elements going into it.

I have one of these built up: viewtopic.php?f=7&t=459&p=6280

You've got me curious to hit the mono button on the monitors.

Re: Summing Left & Right but lowering Center information

Posted: Thu Jan 16, 2014 2:48 am
by vintagedesign
Well mono-ing the signal after a M/S processor with a wide control will just turn your wide control into an expensive level control :D

Re: Summing Left & Right but lowering Center information

Posted: Thu Jan 16, 2014 4:49 am
by mediatechnology
Well mono-ing the signal after a M/S processor with a wide control will just turn your wide control into an expensive level control
Well after a good night's sleep and a cup of coffee that seems to make perfect sense.

Re: Summing Left & Right but lowering Center information

Posted: Thu Jan 16, 2014 5:46 am
by mediatechnology
Is there a simple way to sum L & R and still be able to lower the centre info?
Like to try this in a compressor design with a single sidechain where L&R are summed before the threshold detector.
Maybe a potentiometer?
If L&R are summed ahead of the detector all that you will have is center.
There would no longer be a remaining L or R in the side chain to lower center relative to.
The detector will be deaf to stereo completely.

I don't know what you're ultimately trying to accomplish so bear with me.

Is there some reason that you don't want to sum the output of the detectors then have a single threshold stage?
There might be something here:
viewtopic.php?f=6&t=17
In the above thread I write about how the GSSL sidechain is deaf to stereo but the original SSL bus compressor is "greater of two." (Actually greater of four or eight depending on how you count.)

Why not use multiple detectors?
Off the top of my head it would seem to be useful to have a measurement of correlation/decorrelation.

The correlated signal could be measured by summing audio prior to detection; L+R.
The decorrelated signal could be measured by having an L-R input to a second detector.
Or
The decorrelated components' level differences in L and R could be measured by having individual detectors then performing L-R on the detector outputs.
If a log scale is used then the output would be the ratio of L and R's difference.
The second measurement might not be useful as it would only provide an indication of L and R level differences, not actual decorrelation.

The above measurements of correlation might provide something cool that could be used to change width dynamically or somehow process the signal based on relative energy of the correlated/decorrelated components.