Don't Understand Parallel Compression

can someone explain how this image http://www.soundonsound.com/sos/mar09/images/QA_04.jpg

how does parallel compression work?

if you add two identical signals you get the same thing but louder right?

if you add a signal with a compressed copy of itself, isn’t that still compressing the highs?

Hello,

The two tracks are not the same because one of them has a large amount of compression applied whereas the other is pristine.

Parallel compression is a marvellous technique that allows you to both bake your cake & eat it as well.
It can work in one of 2 ways, depending on your compressor’s controls.
1 - If your compressor has a “mix” control, you can simply insert this to your track and dial in the compression mix you want (more on this later) and it will be combined with the uncompressed signal in the compressor itself.
2 - Create a new Group track in your mixer (I am using the nomenclature from Nuendo as this is my DAW of choice) and insert your compressor as insert #1 (This is my preferred approach, regardless of mix dials, and is the method we will discuss here)

Usage is simplicity itself.
Using EFX Sends, feed the compressor group track with Kick, Snare (maybe hats, but probably not), any toms you want etc. You can adjust this as needed, and with this approach (option 2) you have perfect control over everything.
The idea is that instead of bringing high levels down, we bring quiet levels up - which is much more natural sounding to our ears and the artifacts created by low-level signals make the compression effect less evident in addition to retaining dynamics much better than downwards compression can (downwards compression is the usual approach, and is what you see on your attached plot as “typical insert curve”) because the transients are not being brought down but instead enhanced - you’ll see why when we look at the settings to use…

Initial settings.
Set the attack time as fast as possible to begin with. Release time should be set to taste and results required.
I’d also start with a 10:1 ratio, and again adjust as required.
Using aux sends helps here, as if (for example) we need a bit more oomph on the kick, we simply dial in a few more dB on the send.

I hope this helps to clear it all up a little…

nice explanation. Thanks.

Hi neilwilkes, I know how to use parallel compression, I don’t know how it works.

How is it bringing up low vol instead of compressing high vol? your compressed copy is still compressing the highs and not touching the lows.

Depends on how you set it all up.
Think about it for a while - you still have the original uncompressed group, and because the compressed group is fed by FX Sends, not an ordinary in-line downwards compressor, you are not compressing the highs at all on the original which remains as it was, complete with the dynamics intact.
By compressing in this manner, you are bringing up the quiet parts without hurting the dynamics as you do not have to welly the compressed fader all the way up the track.
Careful tuning of the attack & threshold parameters will - especially if the compressor used has some form of keying or sidechaining (even internal) - raise the quiet parts up or else can be set to give extra oomph to the kick, or more crack to the snare (by adjusting sends) or all at once (by adjusting attack)

The idea is to blend a compressed version with the uncompressed to enhance specific areas only instead of everything. You can even go further than this & use 2 comps in line on the parallel buss, with one set to enhance the punch & the other to enhance the oomph.
It takes practise to get right, but it can be a wonderfully powerful tool.

ya thinking about it some more I got it now

actually I still don’t… oh well

Thanks for the explanation Neil, I’ll give it a try. :slight_smile:

Cheers,
Wim

Honestly, it sounds more complicated than it really is.
Put as simply as I possibly can:
The uncompressed track remains with it’s dynamics fully intact, and is the foundation still.
The compressed track goes (usually) underneath the uncompressed one and depending on how you have the settings, can be either adding the extra oomph at the low end, or emphasizing the attack, or adding some punch.
You can even go further, and use 2 or 3 compressed versions all set to do different things.
Because these tracks are usually underneath the main uncompressed one, things sound a lot bigger as a result than they would with standard insert compression.

Okay, it can be argued that if there is a “mix” control on the buss compressor (or whatever one you use) then this is the same thing, but it does not give you the same level of control.
Your options are enormous - because you’re working with copies of the uncompressed original going through compressors, you can set the EQ slightly differently (try an inverted “loudness” curve on the compressed track) and have all sorts of fun with sidechains.

Hello friends. This is my firt post in this forum. So, please to meet all of you.

Paraller compression, AS I USE IT, is to mix two (or more) differently compressed cloned tracks with the original track.

Special case is when you create only one clone, compress it and mix it with original track, the so-called New York technique.

Every cloned track can be compressed, expanded, de-compressed or de-expanded with different purpouse.

For example one track can be used to de-compress fist atacks and recover dynamics or give punch in the trasients.
Other clone can be compressed (or eq) to avoid high frequency fried sounds.
Other clone can be used to repair holes in the mix, or to reduce the dynamics or wind instruments behind the lead voice, etc.

Then, you can mix in the desired/needed detail the tracks in order to get what you want second after second.

Why we use parallel compression?

Many times you want to expand some sound and compress other ones, but all of them are blended in a bad mix.
Honestly, this is not a creative technique, is a usefull method to have READY TO USE many versions of the original track and then concentrate into the mix of the clones. Of course is dangerous and tedious because you can loose the sense of the music and ends with a worst mix.

Yo can filter the frecuencies in order to get what you want from every clone. You can cut the lowfrecuency in one clones where you made a work thinking in medium frecuencies; in other clone you can cut high and medium parts and then play with the bottom dynamics, etc. It is better to make the clones from original track, name evey one according what we want to do (Low, midle, hi, midle-voice, brake pads, etc). In this way, we will understand better where the focus is when we process every clone. Then filter the clones to isolate the frecuency zones you need to work on them.

Then, process every filtered clone as needed (compressed, expanded, de-compressed or de-expanded).

Do not use paraller compression all the time. Many times you can use Waves C6 or C4, or Flux Alchemist to repair a track.

Of course better is to mix correctly every element of the mix instead to repair a bad mixed song.

At the end, the simply is best, nothing like a discret, light, elegant compression.

Regards !

http://en.wikipedia.org/wiki/Parallel_compression
http://www.mutantaudio.net/tutorials/parallel_compression_tutorial/parallel_compression_tutorial_ma.html

This method is most used to decompress drums. Currently there are other methods to uncompress tracks, but it is very useful to know the basis of this technique.

Regards!

hi all, I still remain a dedicated student of compression and still trying to figure it all out. I have a different question but maybe I’ll just use this thread.

I look at a lot of presets and “typical” settings, and why is the attack time always something like 2ms, 5ms? Doesn’t this allow transients through? (lookahead setting or not)

I understand in analog world it’s not realistic to have 0 attack time but why in digital do they still use 2-5ms attack and let transients through? because of distortion?

I tested it myself and exported many different settings to compare waveforms, and indeed it does create very unnatural transients, so why does everyone do that?

Because transients are the most important part of certain sounds and you don’t want to kill them with compression but enhance them. If after compressing your transients sound unnatural it means that you are over compressing. Or probably you need an attack longer then 2-5ms, not shorter.
Very often I use 10, 13, 30 whatever I need. It depends on the style of music, your taste and how you want to use the compressor.
Maybe you are trying to use a compressor more like a limiter(?).
I have two suggestions:
Don’t use presets.
As a starting point, try to not compress more then 3db.

sorry transient was the wrong word. I mean that say I want to generally smooth something out so I have a pretty long release time, a non-immediate attack lets this “spike” through.

or same thing if I open waves multiband compressor and check a preset the attack is always 2-5ms or something and they must know what they’re doing but I don’t understand why?

Multiband compression is a very different beast to parallel compression, although you could use a multiband comp in parallel if you chose to, I guess.
The best advice is definitely “do not use presets” though.
Instead, learn how to tune a broadband compressor first, before you go anywhere near a multiband compressor, or you will simply not stand a prayer of a chance of getting the best out of it. Honestly.
In general, buss compression is also different to track compression too - you’ll generally be using less of it at a much less aggressive ratio and probably with soft knee too, although that is of course a matter of taste.

I’ve just been learning about parallel compression as well.
I have a handful of drum tracks which I have sent to a ‘Drums’ Group track
What’s the best way to go about setting up parallel compression from here?

Sorry if it sounds like a simple question; I’m still learning my way around Cubase - setting up busses and all that…

Thanks

Depends what drums you want to squash.
If you want to squash the whole lot, you have 2 choices.
A - stick a compressor into the drum group as an insert, adjust according to taste & flavour & adjust levels with the “mix” parameter to allow uncompressed audio through. Or,
B - (I think this is the better approach) create a new group channel, and insert your compressor in this. Then go to your drum group and set up an effect send to this group, and set to (usually) 0.00dB Post Fader.

Alternatively, if you only want to squash the drums & not the overheads, you can send each drum track as an effect send to this subgroup. This gives you a lot more control, but it might not do what you want. It all depends on your taste & what the song says it needs.
For the compressor, I am currently addicted to the UAD FATSO for tape compression rather than normal compression, but any good buss compressor will do the trick - UAD Precision Buss Compressor, SSL G series, URS 1970/1975/1980, UAD Neve 33609, lots out there.
It’s a great little mix trick.

http://www.fabfilter.com/help/ffproc-manual.pdf

Check out Page 14

http://www.youtube.com/watch?v=RkeJqtJGn0I&feature=player_embedded#at=507

And check out 8:18

Great compressor, but the earlyer techniques mentioned here by adding grout tracks and fx sends is also great and works in the same ‘‘Parralel’’ ‘‘New York’’ compression technique.

Hope I haven’t hijacked this topic :blush:

Unfortunately, I only have the standard cubase compressor, so I can’t mix wet/dry; It’ll have to be with group tracks, I think. I need to play around and experiment to hear the differences between compressing the lot and compressing just the drums without OH.


Fabfilters look the business. I watched their vid, and then another, checked out their website… Do any other compressors show the graphics like that? It’s the first I’ve seen like that - though I am pretty new to this side of music!

Hi Paulboy.

There’s absolutely nothing at all wrong with using the stock compressor in a group channel.
You do have the Vintage Compressor too, and there are some great things you can grab from T’InterWeb for free - are you a PeeCee or a Fruit?
If PeeCee I can particularly recommend these:
http://varietyofsound.wordpress.com/downloads/

Just about everything this guy does is seriously usable, but in particular (and with reference to this thread) I can recommend Density MkII & Ferric TDS (works well used as a parallel device to emulate tape compression, which can often be better than audio compression, but that’s a whole new thread altogether…)

Grab the rest whilst you are there, and wonder at the Boot EQ MkII for free…