All times are UTC+01:00




Post new topic  Reply to topic  [ 7 posts ] 
Author Message
PostPosted: Fri Jul 22, 2016 4:10 pm 

Joined: Sun Feb 03, 2013 2:39 pm
Posts: 333
Usually compressors compensate gain changes in roughly constant time, no matter if it's just 1dB or 10dB. So the amount of dB and the speed vary, but the time needed is about constant, which is why we (can) give a timing constant.

However in ST the higher the amount of dB, the longer it takes to compensate. So the amount of dB, the speed and the time needed vary.

So what do the timing constants in ST even mean?
What would I need to do to measure the timing constants, i.e. find any fixed relation?


Top
   
PostPosted: Sat Jul 23, 2016 3:46 pm 

Joined: Sun Feb 03, 2013 2:39 pm
Posts: 333
The plots below show threshold_in_%*(output/input)^(ratio/(1 - ratio)). (threshold, ratio, attack, release) = (-20dB, 2, 50ms, 150ms). The input signals are bursts ranging from -19dB to 0dB in 1dB steps.

The 1st row is ST's analog peak compressor, the 2nd some other compressor. The right column shows normalized versions for easier comparison. The grey lines show the place where the 86% should be reached for the given attack time.

What you see is that ST uses completely different lines for every given input signal. All have different shape and speed, so only one line is close to reaching 86% at the right spot, but this is just a coincidence. All lines should be at least close to the defining point (50ms, 86%), but they're not.
The other compressor behaves extremely regular. All lines are essentially only shifted versions of the same line, so all reach the 86% just fine.

PS: I compressed the x-axis for nicer plotting, so you cannot read the sample values directly from it. Sorry!
Image


Last edited by \_/ on Sat Jul 23, 2016 4:08 pm, edited 1 time in total.

Top
   
PostPosted: Sat Jul 23, 2016 4:07 pm 

Joined: Sun Feb 03, 2013 2:39 pm
Posts: 333
A question for Hans:

If we are far below the threshold and then a 0dB burst happens, it seems ST immediately reacts to it. How can that be?
The attack time should cause a small delay before ST even recognizes that the threshold was exceeded and gain reduction is needed.


Last edited by \_/ on Tue Jul 26, 2016 9:16 am, edited 1 time in total.

Top
   
PostPosted: Mon Jul 25, 2016 9:27 pm 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11187
Hi,

If you take the normalised lines in ST and not only normalise them but also stretch them equally, they should look the same.

What do you mean by a 0 burst? Max level? Or silence? Or something else?


Top
   
PostPosted: Tue Jul 26, 2016 9:14 am 

Joined: Sun Feb 03, 2013 2:39 pm
Posts: 333
Hi, Hans!

Ouch! Yes, stretching them will make them equal, but it isn't supposed to. You just verified a bug.
Stretching makes lines longer. Essentially all your ST lines are just smaller versions of the largest one. It's impossible to give timing constants in this case (or measure them, because they don't exist).

Let's make it simple:

ST: Imagine some vehicle. It uses constant speed 100m per 10s. So 50m takes 5s and 200m takes 20s. Giving a speed is the only thing you can do here. Giving a timing constant like 10s doesn't make any sense, because the length of the way varies. But that's what you do, Hans.

Most other compressors: They take constant time, let's say 10s, no matter if they go 50m, 100m or 200m. To achieve that they must vary speed, so you cannot use that as an indicator. That's why they just give a timing constant, but that' impossible for ST's both analog and digital compressors because it doesn't apply to them. They just don't work that way.

Hans, because you use the normal timing definitions for normal compressors, does this mean you intended to do a normal compressor, too, but made a compressor design error, or is ST the way you wanted it to be and you just confused timing constants and speed?

Because all the ST lines just differ in size, you must have some reference line, like e.g. the one 6dB above the threshold. Which one is the reference line?

The 'zero burst' is a 0dB burst, i.e. maximum level.


Top
   
PostPosted: Tue Jul 26, 2016 10:32 am 
Site Admin
User avatar

Joined: Mon Mar 17, 2008 1:40 am
Posts: 11187
The behavior is what I intended, and the time constant gives how much time it takes to bridge the gap by 86%.


Top
   
PostPosted: Tue Jul 26, 2016 8:57 pm 

Joined: Sun Feb 03, 2013 2:39 pm
Posts: 333
Thanks for the help so far!

I got one step forward and one backward:

The stretching does make the curves the same, so I can read out a unqiue percentage value independent from the input level. However this value does vary with the timing constant given to ST. So it's a timing variable. :lol:
The data I used is from the plots above, i.e. the internal level of the compressor deduced from the input and output signal. Do I need to use different data, like the output magnitude or something (no idea how to make these curves equal though)?


Top
   
Display posts from previous:  Sort by  
Post new topic  Reply to topic  [ 7 posts ] 

All times are UTC+01:00


You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot post attachments in this forum

Search for:
Jump to:  
cron
Powered by phpBB® Forum Software © phpBB Limited