Logic Pro 9 Stupid Compressor question...

Robert Wilson

Logician
... displaying my ignorance with this but here goes:

I was mucking about with the compressor that comes bundled with Logic 9.1.2.

Some strange (to my eyes anyway) results.

I have a vocal which, with no compression - peaks at zero three times.

I apply these settings:

Attack 5.0ms
Release 170ms
Threshold -20dB
Ratio 5:1
Gain 0dB

Now the zero peaks reach -2.3dB, -1.5 and -1.0dB

Change ratio to 3.1:1 (for some reason the compressor's threshold jumps from 2.9:1 to 3.1:1) and the zero peaks reach -2.1dB, -1.5 and -0.9dB

2:1 and the zero peaks reach -1.7dB, -1.2 and -0.8dB

10:1 and the zero peaks reach -1.7dB, -1.5 and -1.0dB

Even at 30:1 I get -2.6dB, -1.5 and -1.0dB.

These results don't make a whole lot of sense to me.

Are these the results that should be expected with these settings?


Cheers,
Robert.
 

Multispace

Logician
Hi Robert,
The peaks you're reading are the peaks that get through uncompressed the first 5 ms (which is your attack setting).
For vocals, you can often set the attack to 0 ms if you want to tame the peaks in your recording.

Hope this helps.

bye,
Mark
 
Upvote 0

Peter Ostry

Administrator
Staff member
The numbers look a bit strange at the first glance but I think nobody can judge the behavior of a compressor just by a couple of peak values. It all depends on the threshold, the dynamics of the signal and the algorithm the compressor uses. Compression curve, if you want. Can be simple or complicated. And is not always logical without specific knowledge of the technical design.

For example, if a compressor pushes the signal down by 3 dB and you make the attack longer, it may well shoot over, because the attack of the signal gets the chance. Changing one parameter can change the result dramatically.

In your example you changed only the ratio but the result depends also on the other parameters. You may be in a situation where another parameter has a great influence and although you turn only one knob the result can be quite different.

If you want to explore the various compressor models you can feed them pink noise through a tremolo (Logic test oscillator and tremolo in the same channelstrip). But this is a lot of work because you got to change everything in the tremolo and in the compressor. You would need a system.


Multispace said:
For vocals, you can often set the attack to 0 ms if you want to tame the peaks in your recording.
Yes, "often" is the correct word, as you say. Not always. The Logic compressor does allow settings which do not fit to the selected compressor model. And some models do not like zero attack. We have to try.

This is not only true for the Logic compressor, btw. Other compressors may also show strange behavior at extreme settings. The developers don't always adapt the settings range to their own taste and this is good because sometimes we want to experiment.
 
Upvote 0

Robert Wilson

Logician
@Mark & Peter:

Thank you both for your excellent replies.
I'm heading off out for the night but will try to have another look at this over the weekend.

Cheers,
Robert.
 
Upvote 0

pfloyd714714

Logician
Having read George's reply, I was wondering if anyone could point to a website or online document which describes the characteristics of the compressor types? To me they're just names without meaning. The manual's description, "The choices are Platinum, Classic A_R, Classic A_U, VCA, FET, and Opto (optical)," is not terribly helpful.

Jim
 
Upvote 0

Robert Wilson

Logician
OK, I've finally found a spare hour to play about with the Logic Compressor again:

@Mark:

Hi Robert,
The peaks you're reading are the peaks that get through uncompressed the first 5 ms (which is your attack setting).
For vocals, you can often set the attack to 0 ms if you want to tame the peaks in your recording.

Thank you. That has indeed helped to tame the peaks, although not completely. Still dabbling.
 
Upvote 0

Robert Wilson

Logician
OK, after some more mucking about I've discovered that changing the compressor type to FET, reducing attack to 0ms, increasing release to 580ms (or anything higher but that's as low as I can set it) and changing ration to 4.2:1 is giving consistent results for the three peaks.

Obviously a lot more to this compression business than I had first assumed. :)
 
Upvote 0

Robert Wilson

Logician
Obviously a lot more to this compression business than I had first assumed. :)
Yeah, dealing with dynamics is quite a mouthful.

Maybe it's time again to mention Mr. Stavrou:
http://www.mixingwithyourmind.com

:whisper: Alone the parts about compressors and limiters are worth the money ...

Yes, I've had that recommended to me before.

Indeed, next time I have some spare cash I may well send it in that direction.

As an aside, I'm listening to a solo'd track in Logic right now. No effects on the channel strip or the master.
Channel strip meter peaks at 0.0, master meter peaks at -5.9. Shouldn't they be the same?
 
Upvote 0

Peter Ostry

Administrator
Staff member
As an aside, I'm listening to a solo'd track in Logic right now. No effects on the channel strip or the master.
Channel strip meter peaks at 0.0, master meter peaks at -5.9. Shouldn't they be the same?
You mean the output channelstrip. The master in Logic is a fader to control all outputs at once. Make sure this fader is at 0 dB.

If the master fader is at 0 dB, your loss of 6 dB sounds as if your pan law is at -3 dB and you have two hops from the original track to the output. With a pan law of -3 dB the signal goes 3 dB down with each channelstrip on the way.

You can either live with it or rise the level in each channelstrip, or set your pan law to "-3 dB compensated". For the latter: File -> Project Settings -> Audio. Then look for "Pan Law".
 
Upvote 0

Robert Wilson

Logician
Thanks for the reply, Peter:

As an aside, I'm listening to a solo'd track in Logic right now. No effects on the channel strip or the master.
Channel strip meter peaks at 0.0, master meter peaks at -5.9. Shouldn't they be the same?
You mean the output channelstrip. The master in Logic is a fader to control all outputs at once. Make sure this fader is at 0 dB.

If the master fader is at 0 dB, your loss of 6 dB sounds as if your pan law is at -3 dB and you have two hops from the original track to the output. With a pan law of -3 dB the signal goes 3 dB down with each channelstrip on the way.

Ahhh.... makes sense now.

You can either live with it or rise the level in each channelstrip, or set your pan law to "-3 dB compensated". For the latter: File -> Project Settings -> Audio. Then look for "Pan Law".

I've been living with it until now so will probably continue to do so. I assume this is the default setting?
I always adjust the output level before final mixdown anyway.

Are there any particular pros and cons of the different Pan Law methods?
 
Upvote 0

Peter Ostry

Administrator
Staff member
I've been living with it until now so will probably continue to do so. I assume this is the default setting?
So far I see the default setting of Logic 9 is "-3 dB compensated". At least this is what a new empty project says and pan law it is a project setting, not global.

If you have the -3 dB pan law by default, this can come from the template(s) you use.


Are there any particular pros and cons of the different Pan Law methods?
Between -3 dB and -3 dB compensated? Not really. With the compensated pan law you can overdrive the next channelstrip when you pan hard, but unless there are sensitive plugins this does no harm. As an advantage you don't have the 3 dB loss at each hop. Therefore I believe that most people use compensated.

Technically, regarding the panning behavior, there is no difference. The extreme sides get 3 dB more which is the purpose of this pan law, otherwise the sides would sound quieter than the middle.

You should of course not change the pan law for an existing mix, this can scew up your levels. You may save your template(s) with another setting if you want to change it.
 
Upvote 0
Top