What Is Parallel Compression?

[oldbobd]

The first in a series of articles on parallel compression by Bob

What Is Parallel Compression?

Parallel compression is the mixing together of two differently compressed signals coming from the same source. One compressor will compress lightly (or not at all) and the other compressor will have more gain reduction.  The heavily compressed signal is usually mixed under (lower in level than) the uncompressed/lightly-compressed signal.  The advantage is that it brings up lower level signals in the mix while leaving the highest signals alone.  It can raise the average loudness of the sound and yet obtain a sound that is quite natural.

Parallel compression is used both in mastering and in mixing.  In mastering it is usually fastest to import the same audio file into two tracks of your DAW or editing program and then processing them differently before mixing them together.  In mixing, you would need to setup two submix aux channels and apply compression differently in each channel before mixing their outputs.

What is Motown Compression?

Motown compression is the use of parallel compression with equalization added before the heavy compression, usually done during mixing.  The equalization is added to accent certain frequencies needed on an instrument (or an entire mix).  Originally it was used for vocals with increased energy at 8 kHZ but with minimal distortion on the tape or disc due to the accented highs - 8 kHz was accented to get the mixes sounding better over AM radio.  The technique however can be used on any instrument in the mix by using a frequency key to that instrument.

What is New York Compression?

New York compression is a drum mixing technique using parallel compression, popular in New York. The heavy compression of the drums tends to be done with high (nearly limiting) compression ratios with the release time set for medium  (about 100 ms.)  Often equalization is applied to the heavily compressed signal. 

What Parameters Do I Use?

A very important thing to remember is that you are compressing and need to base the settings on what sounds good.  Begin (on one channel) by adding a mild amount of compression that sounds good, and a mild amount of EQ if it sounds good.  After you have a good sound, accent the amount of compression (the amount of gain reduction) and EQ a bit more than you really need, but so it still sounds reasonable.  Now mix in this compressed track under the uncompressed track and you will have parallel compression happening for you.

The way to think of parallel compression is that the compressed track should sound good on lower-level information and the uncompressed (lightly compressed) track will sound good on high-level information.

bob

Note:  Read the rules before you post a reply

[Cant Get Enough]

These are the kind of tips and can't find in books. (or at least every book I own) I can mix pretty well, but am clueless about mastering. Only thing my books talk about is you can Eq, limit, compress, etc. etc. Never about techniques though. Can I get a mastering tip a week !? Or is most of this stuff trade secrets !?

[pyrael]

Great tip oldbob! Very informative.

Not that I have data to include, but I have a possible addition to the subject, if you feel that it is warranted for the topic.

I'm quite interested in the ways that parallel compression could be used in applications like de-essing/de-popping and the like. Do you think parallel compression would work in this area, if so, what suggestions would you have?

py

[Silent Bob]

Parallel compression is a way of simulating upward compression without the pumping artefacts.  Things like pops and sibilance need to be supressed by pushing them downwards, not raising something else around it.  There ARE ways of tackling those with traditional compression though.

[oldbobd]

These are the kind of tips and can't find in books. (or at least every book I own) I can mix pretty well, but am clueless about mastering. Only thing my books talk about is you can Eq, limit, compress, etc. etc. Never about techniques though. Can I get a mastering tip a week !? Or is most of this stuff trade secrets !?

Sounds like you should buy my book when it comes out, but in the meantime I'll be posting articles...

I keep no trade secrets - when I tell my secrets people tell me theirs and I learn more  

Take Care

bob

[oldbobd]

Great tip oldbob! Very informative.

Not that I have data to include, but I have a possible addition to the subject, if you feel that it is warranted for the topic.

I'm quite interested in the ways that parallel compression could be used in applications like de-essing/de-popping and the like. Do you think parallel compression would work in this area, if so, what suggestions would you have?

py

What SB said

 

[abtech]

Sidechain EQ in a limiter works very well for sibilance and transient pops that have a defined frequency range (as opposed to random frequencies which are another matter entirely).  The Neve input module has several switching options which allow the equalizer to be placed pre/post or sidechained with the compressor/limiter and/or gate/expander.  I have found several useful applications for these options especially for accentuating or repressing certain frequency based program material.  You can do a lot of interesting reverb shaping by modifying the eq on the returns through the compressor/limiter sidechain.

[Cant Get Enough]

With side chain compression would I insert a Eq turn dowm the frequencies I don't want compressed and increase the frequencies I that I want my compresser to "react" to. Would a multiband compresser ferform the same task?

[Silent Bob]

With side chain compression would I insert a Eq turn dowm the frequencies I don't want compressed and increase the frequencies I that I want my compresser to "react" to

.

You got it.

Would a multiband compresser ferform the same task?

Yes but in a different way.  A wideband compressor with an EQ in the side chain will detect the emphasized frequencies more but when it clamps the signal, it clamps all of it, not just what you emphasized.  A multiband compressor will only react to those frequencies you set.  Deticated Deessers are basically 2-way crossovers with a compressor on the high pass side.  I often use a multiband compressor in mastering to help tame sibilance or overzealous cymbals but I generally set it so that just the high band is reactive while the other bands just sit and twiddle their thumbs.

Now for dealing with pops & stuff like that, I generally prefer to rerecord the problematic vocals.  If I can't do that and it's a digital project, I load the track in Cool Edit Pro.  I set up a script so that if there's a pop somewhere, I just highlight the pop, hit the "P" button and it removes it.  When I'm working in the analogue domain, I'll just switch on a high pass filter for the moment of the pop.

[oldbobd]

With side chain compression would I insert a Eq turn dowm the frequencies I don't want compressed and increase the frequencies I that I want my compresser to "react" to. Would a multiband compresser ferform the same task?

Excellent question.

The answer is not quite.

Boosting 7 kHz in the sidechain will cause the compressor to compress all  frequencies when there is excessive 7 kHz, where a multiband compressoer set to a frequency range of say 6 kHz - 8 kHz would compress just those frequencies when there was exessive energy at 7 kHz.  

I'm not a big fan of multiband compression and I am a fan of sidechain equalization in compression (which is what I assume you mean by sidechain compression).  To my ears, using the sidechain EQ is more natural sounding in most cases.  In Deessing, the offending S is a very narrow band of frequencies and can be gotten under control with very Hi Q sidechain EQ.  It would be interesting to try it out side-by-side using the two techniques although the filter response used in a multiband (relatively gentle slope - low Q)  will automatically mean you could hone in on the offending frequencies better with sidechain EQ.

BTW I'll often opt for regular EQ to handle S's set to very high Q and tuned to the particular voice - sort of the same kind of technique used to control feedback at live gigs. 

Take Care

Bob

 

[beyarecords]

Hi Bob,

Sounds like you should buy my book when it comes out...

Has your book been released yet, and if so where can it be purchase from?

Many thanks in advance.

Beya

[oldbobd]

Hi Bob,

Sounds like you should buy my book when it comes out...

Has your book been released yet, and if so where can it be purchase from?

Many thanks in advance.

Beya

The beta epublication is scheduled for release early July, 2009

You can keep informed about it's progress and other educational serices at my Superdiosc Mastering home page (see link by my signature)

Take Care

bob

[Kevin Wyman]

I can not wait to try this out  I'm a musician trying to learn how to produce better sounding recordings, and tips like these are incredible  Thank you!

[oldbobd]

I can not wait to try this out  I'm a musician trying to learn how to produce better sounding recordings, and tips like these are incredible  Thank you!

You're welcome - you'll find a lot more of these if you search them out

Welcome to the forum.

bob

[ghprod]

in short way,

this like we send and return signal path right? and combine with un compress signal

regards