Parallel Compression Delay timing

[Micah Tolentino]

This is a mastering question as it relates to mastering, but I am using it in the context of my mixing.

I was wondering if some of you can help clarify something I've read in Katz' book "Mastering Audio". Can someone offer me how to properly introduce the the correct time delay for a parallel compression chain. Is there an easier way to do this ITB with PT?

Heres what Im doing:

All things considered assume both channels have a delay of 0 and/or pdc is turned on.


-Route signal to 2 auxillary channels, both will sum at main output.
-Compress 1 signal with an extreme Threshold.
-Convert 5 samples to milliseconds according to my sample rate
-Place a digi delay on second channel and type in converted time in ms.

**Now Katz says the delay is typically 5-10 samples, so what Im doing is recalculating samples to milliseconds by incrementing one at a time, 6, 7, etc..

The book states "it will have 6 dB extra gain when the delay is correct..."

I am stopping this process when it "sounds right" to me.

Heres my questions:


1) Is this the proper way of doing parallel compression as described by Katz?

2) Is there a more efficient way of doing this?

3) Are there anyone using alternate techniques ITB?


Just thought I'd throw that out there. I'd appreciate the input thx.

[Adam Dempsey]

I wouldn't over-think it. In the past I've done this mainly in analogue or in SADiE, but as a starting point just set your two paths up thus:

• Both paths via the compressors - one with its comp bypassed or 1:1 ratio.
• The comp path: at least 2:1 ratio; fast attack; threshold so there's no gain reduction in the quietest sections of the track/signal; release to suit the signal (requires knowing your particular compressor's sound and its "knee").
• Monitoring just the dry (no comp) signal, blend in the comp signal to taste.

= upwards compression... In a nutshell.

[MAzevedo]

Lately, I've been into parallel compression with very, very slow attack times, sometimes as slow as 350ms and very deep thresholds so the comp is always in gain reduction. Fills the quite bits out and punches up transients all at once. Just something I'm digging folks might want to try.

In regards to the original question, sticking a bypassed or flat ratio instance of the same plug into the "dry" signal is probably the easiest way to match the channel delays.

[masteringhouse]

Originally Posted by Micah Tolentino

I was wondering if some of you can help clarify something I've read in Katz' book "Mastering Audio". Can someone offer me how to properly introduce the the correct time delay for a parallel compression chain. Is there an easier way to do this ITB with PT?

The easiest way to do this in PT is to insert the same plug-in in the uncompressed track and bypass them. This inserts the same latency to both aux tracks even though they are not being used. If using an HD system you can also use the time adjuster plug to add sample accurate delay (verify the delay times in the track).

[beyarecords]

Originally Posted by Micah Tolentino

3) Are there anyone using alternate techniques ITB?

From an ITB perpective, and if using an application like Cubase for example, one would:

1. Create your track
2. EQ your track (If required)
3. Create an FX Channel and assign a compressor to it (plug-in or outboard gear).
4. Go to your original track and choose your newly created FX Channel as a send.
5. Set that send to pre-fader and set the fader to unity.
6. Adjust the compressor on your FX Channel to taste.
7. Mult the two signals together.

Possible compressor settings could be:

Threshold: -50db to -60db
Ratio: 2.0:1 to 2.5:1
Attack: 1ms (take the smallest value possible!)
Release: 250ms
Makeup: 0dB

and then mix it back to the other 'clean' channel from -20dB upwards.

Of course the application already compensates for any delay, so no issue there.

HTH's

[PatrikT]

Originally Posted by Micah Tolentino

2) Is there a more efficient way of doing this?

To not run things parallel at all.

Highly efficient, reliable and right in time every time.


BRGRDS
Patrik

[FBM]

Wow PatrikT! You are a great help!
Thanks Dtuttvid

[MAzevedo]

Originally Posted by PatrikT

To not run things parallel at all.

Highly efficient, reliable and right in time every time.


BRGRDS
Patrik

I have lots of tracks where parallel comp is 'right' and straight comp isn't. I have some tracks where M/S comp is right. Once in a while, I get a track where upward expansion is right. Sometimes, no comp at all is right. Given the wonderful variety of music in the world, I can't think of anything that is right _every_ time.

[Lagerfeldt]

If you have Logic 8 the new compressor will allow parallel compression without using any additional tricks.

Simply click the extended parameters triangle button, and set the mix % slider.

Considering the new compressor engines (e.g. VCA, FET or OPTO) are pretty good, that's a great option for doing it ITB.

[alexhitmakers]

Originally Posted by masteringhouse

The easiest way to do this in PT is to insert the same plug-in in the uncompressed track and bypass them. This inserts the same latency to both aux tracks even though they are not being used. If using an HD system you can also use the time adjuster plug to add sample accurate delay (verify the delay times in the track).

i notice that sometime if you bypass the plug it will change the delay

i don t use PT so i don t know with this DAW but with many other i experienced such a problem

i always use a ratio of 1:1 instead of the bypass

alex

[PatrikT]

Originally Posted by MAzevedo

Given the wonderful variety of music in the world, I can't think of anything that is right _every_ time.

I think real time is right every time.

How can a reconstruction of a mix (the paralell signal), when it is fairly aligned with the original cause anything else than diffusion to the original mix anyway?


BRGRDS
Patrik

[FBM]

Hi PatrikT,

Bob Katz wrote this about it:
"The parallel compression technique employs a matched time delay in the "dry" signal path to avoid phase shift or comb filtering. This yields very transparant-sounding upward compression"
Read his book and learn something!

Cheers David

[MAzevedo]

Originally Posted by PatrikT

I think real time is right every time.

How can a reconstruction of a mix (the paralell signal), when it is fairly aligned with the original cause anything else than diffusion to the original mix anyway?


BRGRDS
Patrik

You can also do parallel compression fully in the analog domain, which results in no delay to compensate for. Several analog compressors have a "Mix" control for this purpose. There are also a number of plug-in comps that have an included mix feature. You should give it a listen some time, it is a very, very useful technique, and you are selling yourself short if you don't at least give it a listen and see how you can use it.

[dcollins]

Originally Posted by MAzevedo

You should give it a listen some time, it is a very, very useful technique, and you are selling yourself short if you don't at least give it a listen and see how you can use it.

I've never had any luck with parallel compression either. In an all-analog chain with a wet/dry balance control.

It's denser, yes, but murkier and less good than just regular compression, imo.

DC

[peeder]

Delay Compensation in Pro Tools - FREE Video Tutorials!

I use parallel compression often and am able to get it to work ITB and OTB...

[uros]

Hi Micah Tolentino,

One way you can find a delay time is: Make the copy of the track, insert a compressor plugin in it and leave the compressor settings "flat" (ratio 1:1, treshold 0...). Then you can reverse the polarity on the 1'st (unprocessed) track and adjust it's time delay until tracks cancel each other.

Hope this helps,

Uros

[j-madd]

I run cubase sx 3 and tried parallel compression for fun. As suggested in Mr. Katz' book I copied the track and alligned, inserted compressor on copied track, then inverted the phase on 1 track to see if it would null out. While the volume did dip substantially when playing both channels, it did not fully go away, making me think there is still a misallignment/latency issue. Cubase is supposed to have delay compensation, but this would cause me to question the accuracy.

Justin

[Micah Tolentino]

Thanks for all of the replys so far. Its great to know how each of you are doing this with the different DAWS as to see perspective should I need to mix on your platforms in the future.


Uros,

Thanks for pointing that out. I understand that the non-compressed track can be inverted, and nudged to null, but I was wondering how this was done using aux and busing as opposed to physically making another track. In the instance that I need to parallel comp a drum buss, or something that is composed of a mult (say a 4 part harmony), making a copy is inefficient as you would need to re-record the copy everytime you edit the original (take down a snare, tenor, etc...)


Peeder, I think this is exactly what I was looking for here:

MDC05-NullingMethod


This showed me exactly how to invert the track and nudge with a digi plugin. It can be applied to the aux track without having to make a physical copy. In this example the guy compensated his delay for his outboard FX using 1 plugin. Perfect. Thanks

[peeder]

Originally Posted by Micah Tolentino

Peeder, I think this is exactly what I was looking for here:

MDC05-NullingMethod


This showed me exactly how to invert the track and nudge with a digi plugin. It can be applied to the aux track without having to make a physical copy. In this example the guy compensated his delay for his outboard FX using 1 plugin. Perfect. Thanks

Happy to be of service. Note that with analog i/o you will also need to adjust level to null as much as possible (the converters will prevent an absolutely perfect null because of their limitations of course). A trim plugin can help adjust level by clicking on the gain field and nudging the value up and down with the arrow keys.

[Micah Tolentino]

Sorry, I didnt realize that "guy" was you Peeder until I backtracked to chapter 1. Great stuff, and I agree whether you are on HD or LE, or whatever, you need to know this stuff regardless, especially if you are using inserts or mixing outboard with ITB (which I do).

Great wealth of knowledge there. Thanks for sharing.

[thethrillfactor]

Originally Posted by Micah Tolentino

Thanks for all of the replys so far. Its great to know how each of you are doing this with the different DAWS as to see perspective should I need to mix on your platforms in the future.


Uros,

Thanks for pointing that out. I understand that the non-compressed track can be inverted, and nudged to null, but I was wondering how this was done using aux and busing as opposed to physically making another track. In the instance that I need to parallel comp a drum buss, or something that is composed of a mult (say a 4 part harmony), making a copy is inefficient as you would need to re-record the copy everytime you edit the original (take down a snare, tenor, etc...)


Peeder, I think this is exactly what I was looking for here:

MDC05-NullingMethod


This showed me exactly how to invert the track and nudge with a digi plugin. It can be applied to the aux track without having to make a physical copy. In this example the guy compensated his delay for his outboard FX using 1 plugin. Perfect. Thanks

If you are trying to do parallel processing by doing a roundtrup with converters you will never get it to completely null.

Also there will be some phase shifts with peaks and nulls.

How much it affects the audio depends on the sampling rate.

Parallel processing with analog gear is best done in the analog domain.

With plug ins its a different story.

[uros]

Originally Posted by Micah Tolentino

Thanks for all of the replys so far. Its great to know how each of you are doing this with the different DAWS as to see perspective should I need to mix on your platforms in the future.


Uros,

Thanks for pointing that out. I understand that the non-compressed track can be inverted, and nudged to null, but I was wondering how this was done using aux and busing as opposed to physically making another track. In the instance that I need to parallel comp a drum buss, or something that is composed of a mult (say a 4 part harmony), making a copy is inefficient as you would need to re-record the copy everytime you edit the original (take down a snare, tenor, etc...)

Dear Micah Tolentino,

Steinberg's Nuendo and Cubase (I am not sure about PT) have a polarity invert switch on an FX Track (receive track), so you can do the null test in the same way as in the previous example (with a copy of the track). Of course, this should work as far as you stay ITB.

Hope this helps,

Uros

[peeder]

Here's an audio example of what can be done with parallel compression analog-on-digital.

http://www.gearslutz.com/board/2080560-post69.html

It was done at 44.1, yes you might do slightly better at 96, but I don't make any apologies for the end result, which I think is more than fine.

[masteringhouse]

I ran a few null tests with parallel comp using the bypass technique and no bypassed plug using PT delay compensation. My results were that as long as the amount of latency delay is less than the amount specified in the delay compensation preferences of PT it will null fine. However if the amount of delay introduced by a plug is greater than that value it will not and requires that you delay in another manner (bypassed plug works fine). Some plugs can have larger values that what is allowed for in the preferences, or if you use multiple plugs the sum can be greater. Using bypass takes the guesswork out.

[j-madd]

Great Idea, put the compressor on both tracks and bypass one. This will take care of having to worry if latency compensation is actually working.

Justin

[DozerMayne]

Doesn't PSP's Mastercomp "Mix" knob give you the same thing, as Bob Katz method?

[peeder]

Originally Posted by DozerMayne

Doesn't PSP's Mastercomp "Mix" knob give you the same thing, as Bob Katz method?

I'd think so...