The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
Parallel Compression: Haven't found the answers
Old 12th April 2012
  #1
Gear addict
 
NotchontheRocks's Avatar
 

Thread Starter
Parallel Compression: Haven't found the answers

I just had a few questions about use of parallel compression while mastering. I understand what it does and how to accomplish it, but when does one determine when it is appropriate to use? What aspects of a recording/song do you hear that make you go to parallel compression? I understand the whole "do it if it sounds good," but downward/upward compression and parallel are pretty different and might be time consuming for trial by fire. Which leads me to my next question...is it ever appropriate to use multiple types of compression (downward and parallel)?

When and why do you choose to go to one type of compression or another?
Old 12th April 2012
  #2
Lives for gear
 
Jerry Tubb's Avatar
 

Verified Member
Quote:
Originally Posted by NotchontheRocks View Post
I just had a few questions about use of parallel compression while mastering. I understand what it does and how to accomplish it, but when does one determine when it is appropriate to use? What aspects of a recording/song do you hear that make you go to parallel compression? I understand the whole "do it if it sounds good," but downward/upward compression and parallel are pretty different and might be time consuming for trial by fire. Which leads me to my next question...is it ever appropriate to use multiple types of compression (downward and parallel)?

When and why do you choose to go to one type of compression or another?
The Rupert Neve Master Bus Processor does a really nice job of PC.

It's the knob labelled "Blend".

JT
Old 12th April 2012
  #3
Lives for gear
 
Ienjoyaudio's Avatar
 

I find the best time to use parallel compression for a mastering application is when im asked to achieve absolute maximum loudness. Usually around -10 to -8DB RMS (using a -18 scale) Blending the two different signals greatly increases the RMS value. Using multiple types of compression is just fine. Adding limiters after the parallel processing is a must.
Old 12th April 2012
  #4
Lives for gear
 
sat159p1's Avatar
It's good to keep the transients alive
Old 12th April 2012
  #5
Lives for gear
 

Quote:
Originally Posted by NotchontheRocks View Post
downward/upward compression and parallel are pretty different
No, parallel compression = downward compression.

Mixing a compressed signal with a dry signal is the same as lowering the compression-ratio.
Old 12th April 2012
  #6
Lives for gear
 
Adam Dempsey's Avatar
 

Verified Member
Series compression = regular downwards compression.
The net effect of parallel compression is upwards compression, in that below threshold the two signals merely sum, and above threshold the final output comprises increasingly less compressed signal & hence more uncompressed signal.

References to 1977 description here: http://homepages.paradise.net.nz/~rh.../sidechain.htm

Last edited by Adam Dempsey; 16th May 2012 at 01:51 AM.. Reason: added 1977 reference
Old 12th April 2012
  #7
Lives for gear
 

Quote:
Originally Posted by Adam Dempsey View Post
Series compression = regular downwards compression.
The net effect of parallel compression is upwards compression, in that below threshold the two signals merely sum, and above threshold the final output comprises increasingly less compressed signal & hence more uncompressed signal.
That's a common wrong way of imaging what it does. There's actualy more compressed above the threshold, because below threshold, there's no compressed signal at all (cause even the compressor signal is not compressed)

Since upward compression looks like this


and downward compression looks like this


parallel compression is NOT upward compression but downward. I don't know who spread this false information, you can read it even in wikipedia , but it's wrong. Generate a simple ramp and see what it looks like after parallel compressing it.
Old 12th April 2012
  #8
Gear Addict
 
Joelistics's Avatar
 

Quote:
Originally Posted by evosilica View Post
That's a common wrong way of imaging what it does. There's actualy more compressed above the threshold, because below threshold, there's no compressed signal at all (cause even the compressor signal is not compressed)

Since upward compression looks like this


and downward compression looks like this


parallel compression is NOT upward compression but downward. I don't know who spread this false information, you can read it even in wikipedia , but it's wrong. Generate a simple ramp and see what it looks like after parallel compressing it.
Maybe the diagrams would be more relevant if you could draw one for parallel compression too, as it is (or at least can be) very different to regular downward compression and in many (or most) cases the same as upward compression particularly when using fast attack times low threshold, mixing the wet signal quietly under the dry signal. Thereby you effectively only work on lifting low level information.
Old 12th April 2012
  #9
Lives for gear
 

Quote:
Originally Posted by Joelistics View Post
as it is (or at least can be) very different to regular downward compression
no, it's not

Quote:
Originally Posted by Joelistics View Post
Thereby you effectively only work on lifting low level information.
That's the common wrong way to imagine how it works - i don't know who came up with that. It's lifting the lower levels just as any compressor with makupgain lifts the lower levels - that still doesn't make it upward!! it's still the higher levels, that are more compressed - parallel or not.
believe me or not or even better, try it out, feed your compressor with tone ramps and see, that it's not upward at all :D

Quote:
Originally Posted by Joelistics View Post
Maybe the diagrams would be more relevant if you could draw one for parallel compression too
Here you go:

infinity:1 mixed with dry by 50:50 results in 2:1 compression! Same threshold! Just reduced ratio.


Old 12th April 2012
  #10
Lives for gear
 
Greg Reierson's Avatar
 

Verified Member
I use parallel compression to fill in gaps. For example, if there is low level ambient sound that needs to come up it can be great. It can also be good for leveling out an overly dynamic mix. I use it maybe once a month on a song or two.

GR
Old 12th April 2012
  #11
Gear maniac
 

Evosilica, you're close, but it doesn't quite work as you suggest.

If you take two signals, even if the highly compressed signal drops totally out to no signal past the threshold, the most your signal can possibly decrease is -3db if they are mixed evenly. The remaining dry signal without pumping effects will dominate above the threshold because it is louder and will mask the squashed signal.

Below the threshold, the gain-compensated compressed signal will dominate, as the low-level content can be raised as much as 10 or 20db compared with the dry signal, and will greatly dominate until the threshold. Even at these great levels, you are not affecting the high level ratio much at all.

The difference, IMO is that the majority of the signal you are hearing is coming from the uncompressed parts -- the below-threshold part of the compressed signal, and the aove-threshold part of the dry signal. The greater the compressor works, the LESS you hear it at the high level signals as you are then left almost totally with the dry signal. Compare with typical compression where the greater the compressor works, the MORE you hear it as it is the only thing available.

Google Parallel compression SOS and you can find Sound On Sound's article that details the process and gives a parallel compession curve compared with standard compression curve.

I would suggest the technique for when you are trying to bring up low level content but typical compression is affecting the high level content too greatly.

via Gearslutz Mobile App
Old 12th April 2012
  #12
Old 12th April 2012
  #13
Lives for gear
 

It's funny, that noone belives me :D

Sorry, but there's no way to get the resulting transfer function, that is posted on SOS with parallel compression. I'm shocked they publish sh1t like that. A case of poor research + "experienced audio "engineers" say so, so it's gotta be right", i guess.
In contrast i have actually measured parallel compression setups with different settings.

Can't you just try it?? Just feed a simple tone ramp through your parallel compressor setup to see, that there's no way to get that behaviour?

Quote:
Originally Posted by RScott View Post
Below the threshold, the gain-compensated compressed signal will dominate
Below the threshold, there is NO compression going on. That's what a threshold is good for. So that alone proves, it's downward!!!

Quote:
Originally Posted by RScott View Post
If you take two signals, even if the highly compressed signal drops totally out
That's a super special non practical case, like ratio -1:1 or something. And still it doesn't make it upward. Please google what upward actually does. It alters dynamics below threshold. This can never happen with parallel compression when using a downward compressor. Why is this so hard to understand??


In a practical scenario, you would set the ratio like 10:1 and the threshold so low, that the compressor is always compressing. Then mix like 10:90 or 20:80---

This results in a transfer function as you would take a compressor with very low threshold and like 1.1:1 or 1.3:1 .... which is - oh suprise - also a common mastering compression technique (if the compressor lets you dial in such small ratios)

still don't believe me??
Old 12th April 2012
  #14
Lives for gear
 
Lenzo's Avatar
I think you'd be better off just putting in the time and decide what you like and how you like to do it. You seem to have enough knowledge. Ask 20 people, get 20 different opinions...blah, blah, blah.
L.
Old 12th April 2012
  #15
Lives for gear
 
Red Mastering's Avatar
 

Verified Member
Quote:
Originally Posted by evosilica View Post
It's funny, that noone belives me :D

Sorry, but there's no way to get the resulting transfer function, that is posted on SOS with parallel compression. I'm shocked they publish sh1t like that. A case of poor research + "experienced audio "engineers" say so, so it's gotta be right", i guess.
In contrast i have actually measured parallel compression setups with different settings.

Can't you just try it?? Just feed a simple tone ramp through your parallel compressor setup to see, that there's no way to get that behaviour?



Below the threshold, there is NO compression going on. That's what a threshold is good for. So that alone proves, it's downward!!!



That's a super special non practical case, like ratio -1:1 or something. And still it doesn't make it upward. Please google what upward actually does. It alters dynamics below threshold. This can never happen with parallel compression when using a downward compressor. Why is this so hard to understand??


In a practical scenario, you would set the ratio like 10:1 and the threshold so low, that the compressor is always compressing. Then mix like 10:90 or 20:80---

This results in a transfer function as you would take a compressor with very low threshold and like 1.1:1 or 1.3:1 .... which is - oh suprise - also a common mastering compression technique (if the compressor lets you dial in such small ratios)

still don't believe me??
it would be downwards if you wouldn't mix it with dry signal,
now proportions of dry/wet could change this, when 20 wet to 80 dry, you got upward (imo), and when add more wet, on some point it's downward...
anyway I am just testing my new compressor P38EX tfpro,
who has wet/dry mix knob,
to the OP = it's not time consuming, simply dry/wet mix knob can do wonders,
different wet/dry combinations with different threshold/ratio can give you loads of options,
I am learning sweet spots of my new comp,
already noticed ratio 1.2:1, slowish att, fast rel., threshold - shaving 2-4dBs,
and wet dry mix - 75/25, or 80/20 - sounds sweet, transparent and smoothing&gluing mix very nice
I tried aggressive compression (1176 mode or sc2 mode) and then mix it only 20%wet to 80% dry - great!
that little mix knob is a great tool!
I am getting TK AUDIO Blender soon, to have another 2 extra wet/dry knobs!
and utilize it with another compressor and Culture Vulture
Old 12th April 2012
  #16
Lives for gear
 

Quote:
Originally Posted by Red Mastering View Post
it would be downwards if you wouldn't mix it with dry signal,
now proportions of dry/wet could change this, when 20 wet to 80 dry, you got upward (imo), and when add more wet, on some point it's downward...
just test it out to prove yourself wrong. just because it's a common opinion doesn't mean it's true. it was a common opinion, that the earth is flat once.

but since you all don't care what i've said and you don't test it yourself, i did it for you.
i can't believe i have invested this much time

if you can't see the difference between parallel and upward and the similarities in parallel and low ratio compression then i don't know

SORRY FOR HUGE POST

1. No Compression



2. Downward Compression [2:1, Thresh=-15dB, 30ms Attack, 300ms Release]



3. Upward Compression [2:1, Thresh=-15dB, 30ms Attack, 300ms Release]


4. Parallel Compression [80dry:20wet, 2:1, Thresh=-15dB, 30ms Attack, 300ms Release]


5. Downward Compression [1.1:1, Thresh=-15dB, 30ms Attack, 300ms Release]


cheers
Old 12th April 2012
  #17
Lives for gear
 
Red Mastering's Avatar
 

Verified Member
it seems like you misinterpreted it,
on your screen parallel looks quite different then downward and upward,
also proportions of wet dry would change it (that's just my opinion and I could be wrong)
I don't claim I KNOW IT, and SOS article and all of you are dumb fuc"*&s)
I am happy to be proven wrong and learn something
Old 12th April 2012
  #18
Lives for gear
 

Quote:
Originally Posted by Red Mastering View Post
it seems like you misinterpreted it,
on your screen parallel looks quite different then downward and upward
what did i misinterpret?

and where do they look quite different?
parallel [4] und low-ratio [5] behave almost identical. If that's not apparent, i give up. sorry guys .....
Old 12th April 2012
  #19
Lives for gear
 
Red Mastering's Avatar
 

Verified Member
attack seems similar but decay is different, isn't it ?
don't give up too early,
now it's too late))
Old 12th April 2012
  #20
Gear Addict
 
Joelistics's Avatar
 

Quote:
Originally Posted by evosilica View Post
just test it out to prove yourself wrong. just because it's a common opinion doesn't mean it's true. it was a common opinion, that the earth is flat once.

but since you all don't care what i've said and you don't test it yourself, i did it for you.
i can't believe i have invested this much time

if you can't see the difference between parallel and upward and the similarities in parallel and low ratio compression then i don't know

SORRY FOR HUGE POST


cheers

Just using the same settings on both parallel and upward compressors isn't going to get you there. They are different process that if used appropriately can yield the same results. The parallel compression in your example should read something like this to be similar to the upward compression:

threshold: -50db
preview: 20 ms
attack: 0.1 ms
release: medium (depends really)
ratio: 2:1

dry: 70%
wet:30%

Notice how the peaks in this example remain unchanged?
Old 12th April 2012
  #21
Lives for gear
 

Please, just show me one "self made" graph (not the wrong one from SOS or something), where parallel compression results in a upward-compression-style transfer-function. You know, upward - levels below threshold are increased, levels above threshold are unchanged. I can send you the wav i've used for testing, if you want.

No matter which settings you'll choose, this will never happen!! It will always be a downward style function. Period.


Until that i see no sense to post anything in this threat, as i've made my point and noone has really proven me wrong yet.
Old 12th April 2012
  #22
Lives for gear
 

Quote:
Originally Posted by Red Mastering View Post
attack seems similar but decay is different, isn't it ?
don't give up too early,
now it's too late))
I don't see a dramatic change in time-comstants, but that's not the main point. The main point is, the resulting function is clearly downward! Gain is reduced above threshold, not increased below, as upward does!
Old 12th April 2012
  #23
Lives for gear
 
Flying_Dutchman's Avatar
 

imo:
the function is downward - well the process is downward, but it´s kind of upward fake...

if you choose high ratio/low threshold/high GR/ fast attack for the parallel process it´s kind of upward if you blend them, cause you´re bringing in the valleys rather than reducing the hills
levels below the treshold aren´t increased technically, but the levels above are reduced dramatically and if you blend them in the "below threshold" signals get more...depend on the release of course

the fx is nice, cause you don´t kill all transients with high gain reduction...if compression is the way to go
peace
Old 12th April 2012
  #24
Lives for gear
 
Greg Reierson's Avatar
 

Verified Member
Quote:
Originally Posted by evosilica View Post
levels below threshold are increased, levels above threshold are unchanged...

No matter which settings you'll choose, this will never happen!!
Yep, that's exactly how it works. The main signal is unaffected. When mixed, the parallel channel adds significantly when under threshold (because it's not being compressed) and adds little above threshold (because it is being heavily compressed with a high ratio, fast attack, slow release and 10-2-dB of GR during the loud spots). That is how it lifts low level detail but leaves loud sections mostly as they were.

Quote:
Originally Posted by evosilica View Post
Until that i see no sense to post anything in this threat, as i've made my point and noone has really proven me wrong yet.
You are wrong. My guess is that you don't understand how to properly set up a parallel compression path.


GR
Old 12th April 2012
  #25
Lives for gear
 

Quote:
Originally Posted by Flying_Dutchman View Post
imo:
the function is downward - well the process is downward, but it´s kind of upward fake...
i agree on fake upward, because it sound quite similar as subtle upward, but then you could also call "low threshold - low ratio (1.1-1.2:1)"-downward-compression "fake upward", as it has the exact same transfer function.

Quote:
Originally Posted by Flying_Dutchman View Post
it´s kind of upward if you blend them, cause you´re bringing in the valleys rather than reducing the hills
It seems logical, that it brings up the valleys, but it doesn't do it technically. It's still the hills, where more effective gain reduction is happening = downward

i am ready to be proven wrong by a self measured graph though heh
Old 12th April 2012
  #26
Lives for gear
 

Quote:
Originally Posted by Greg Reierson View Post
Yep, that's exactly how it works. The main signal is unaffected. When mixed, the parallel channel adds significantly when under threshold (because it's not being compressed) and adds little above threshold (because it is being heavily compressed with a high ratio, fast attack, slow release and 10-2-dB of GR during the loud spots). That is how it lifts low level detail but leaves loud sections mostly as they were.
LOL, it's easy to say you are wrong, but not to prove me wrong. show me the setting, that results in an upward-style graph. i have attached my test tone. just feed it through your settings and see it never behaves like upwardcompression!

Quote:
Originally Posted by Greg Reierson View Post
My guess is that you don't understand how to properly set up a parallel path.
LOL
And my guess is you will never show me an actual prove, that i'm wrong. thought it would be so easy, if you were right
Attached Files

Compression-TestTone.wav (689.1 KB, 261 views)

Old 12th April 2012
  #27
Lives for gear
 
Laurend's Avatar
 

The parallel compression (0 attack, infinite ratio no make up gain) allows a 6 dB gain for levels below the threshold without touching the attack envelop of the signal.
It's a different beast than the upward compression as shown in this video.
Compressor vs De-Expander - YouTube
If you need a 9 dB boost for low levels, just add a second compressor in parallel.
Old 12th April 2012
  #28
Lives for gear
 
Greg Reierson's Avatar
 

Verified Member
Here's a quick test. Two sound files and two images. Before and after. This is clearly an bottom - up form of gain manipulation where the loud section is mostly untouched and the quiet section is increased in level. The parallel path was about 3:1 with a fast attack, a slow release with a hefty amount of gain reduction during the loud sections and much less during the quiet sections. You can see the recovery in the quiet spots. Obviously, it would have to be tweaked for audio content but this is the general idea of how it works.

GR
Attached Images
File Type: jpg Source Image.jpg (31.1 KB, 436 views) File Type: jpg Result Image.jpg (30.0 KB, 426 views)
Attached Files

Parallel Comp Source File.mp3 (321.0 KB, 482 views)

Parallel Comp Result File.mp3 (321.0 KB, 533 views)

Old 12th April 2012
  #29
Registered User
 

From a mixing perspective I'm generally in your camp Evo. Parallel compression is largely overrated.

There are quite obvious reasons why it doesn't sound and isn't the same though. The difference isn't huge in real life applications and that is largely why I don't really care for it but there are differences.

You should try experimenting with impulses rather than simple tones. the differences come from the transient response.

Try some drums and flatten them with an L2 or similar and leave that sitting at -10. Add back in the dry uncompressed drums peaking at 0.

Try and emulate that exact sound with a single compressor. You just cant.
Old 12th April 2012
  #30
Lives for gear
 
recordinghopkins's Avatar
 

Verified Member
evosilica gets points from me for using cool edit pro!
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Similar Threads
Thread
Thread Starter / Forum
Replies
illiummusic / So much gear, so little time
7
damien / So much gear, so little time
3
mic anon / So much gear, so little time
24
AlexLakis / So much gear, so little time
6
FFTT / Geekslutz forum
43

Forum Jump
Forum Jump