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 under microscope Dynamics Plugins
Old 18th August 2008
  #31
Lives for gear
 
Nordenstam's Avatar
 

Verified Member
Hi Vitaly!

You are 90% there in making a fine paper on parallel. A bit more work and it'll shine! As a serious infojunkie, I would really appreciate if you can find some time to edit the article. The information is not totally correct as is. Hope you update the thread on that other forum as well. Linking to the (from what I've gathered, original, 1996) paper by Richard Hulse would be appropriate, IMHO. The calculations in the appendix are useful too as they express the amount off added boost in dB. More important is the general guidelines for the effects of threshold, ratio and gain on the parallel transfer curve at the bottom of that page. These interactions are not included in your article as far as I could tell. A few such corrections to your article and it'll be great!

The problem is not that you've found a way to calculate equivalent values for parallel and regular compression. That's cool. It's the conclusions that are misleading. Figure 2-5 and the text around them gives the impression that there is no difference between the ways of using compression. Figure 6 and 7 are better because they do at least show that there are indeed something going on. Though figure 6 can be understood in many ways, the red stippled line is un described and the curves seems to be the opposite of figure 7. The shape of the parallel curve in fig. 7 seems a bit odd for a real world compressor too. But there are many odd curves to be found when fooling around with the settings, so it may be possible for all I know. Are they measured? Perhaps you should post the settings used too? If the transfer curves are to describe the general trend of using (fast attack slow release) parallel, they should look more like the ones I posted above.

There's nowhere in the text that I can find stating that parallel is exactly the same as regular. But you do use all of the article to allude to that point, using the Bob Katz example to that effect, and so on. The main point seems to be that it's only a question of matching ratios. As it stands now, that's what the end user will probably read. Is there anything positive to gain from that? It would be sad if they didn't bother to learn to use parallel because of an article(seems unlikely, but they may be put off searching deeper..).

Another point: Distortions may come from shaping the compressor response, but there is also inherent distortions in the actual units being used. A parallel comp mixed in at -12dB will also have 12 dB less of that distortion/noise. So parallel scheme does indeed lower some of the distortion. And lastly, the "hybrid curve" is not a mix of logarithm and linear, it's a mix of the compressor transfer curve and linear.


Best regards,

Andreas Nordenstam
Old 19th August 2008
  #32
Lives for gear
 

Quote:
Originally Posted by Vitaly View Post
bmanic, where I called it 'near identical'? I said: 'There is a very small difference. Whether it is worth setting up the parallel scheme is totally up to you.'

BTW I'm happy with my Genelec. Thanks.

Guys, apparently you've got it wrong. Nobody is going to take Parallel compression from you My goal was to share the results I've come to as I thought it could be interesting to someone else. There is no need for anger here.

Kind Regards,
Vitaly.
I think your work is admirable and it explains the issues to people well. But I think in reality the concept is far simpler than this discussion seems to be indicating? And no - it cannot ever be the same thing as pushing the same track through a single compressed path - unless the compressor used is devised deliberately to mimic parallel compression. I don't know of any off-hand.

When you compress a track and mix it with the original what you are really doing quite simply is allowing the compressed path to bring up the levels in the quieter bits when the direct path is not loud enough.

But most importantly - you are allowing the direct (uncompressed) path to dominate for loud passages and events.

So this means that for loud punchy events you are mostly listening to the original signal - and filling in the 'gaps' with the compressed stuff which is less prominent and normally not a major signature in the character of rock tracks etc.

Quite apart from the input/output gain characteristics of this - which you cover well - the major reasons this will sound different to setting the same input/output curve for a single comp over the whole signal are:

1. The contribution of the compressor to the loud and transient events is much less - so the timing characteristics of the compressor are much less of an issue.

2. The compressed path can have a wider range of timing settings without ruining the impact of the loud events - so you can do slower or faster settings without ruining the impression of the whole thing.

3. Therefore a far less 'capable' compressor can be used - simply because it's major contribution to the sound mostly doesn't happen during the impact-defining events. :-)


The biggest downside to compressing this way on a master buss is that it does not control the peak levels - and this is a big issue in a world where the 'red light' has become everything..
Old 19th August 2008
  #33
Audio Alchemist
 
Lagerfeldt's Avatar
Paul, you are the man.

Regarding that last point, I think most people use parallel compression followed by traditional downward compression and peak limiting. So parallel compression simply becomes another tool in the arsenal.
Old 19th August 2008
  #34
Lives for gear
 

Quote:
Originally Posted by Lagerfeldt View Post
Paul, you are the man.

Regarding that last point, I think most people use parallel compression followed by traditional downward compression and peak limiting. So parallel compression simply becomes another tool in the arsenal.
Yes I agree - this is necessary for bus applications.

The problem with comp/limiting is that the need for punch conflicts badly with the need for fast attacks to avoid peak overloads in hyper-loud production styles.

When you limit the peaks you are trading the original signal overshoot for a splash of harmonic distortion instead. This may sound louder and produce more apparent level, but it may not be what you would prefer artistically, as higher order harmonics sometimes tend to sound tinny and weak. So the actual harmonic content of limiter overdrive is really important to the conception of limiter design.

It's a conundrum :-(
Old 20th August 2008
  #35
Audio Alchemist
 
Lagerfeldt's Avatar
It's a compromise.
Old 27th August 2008
  #36
Can a question ever be silly?

What is it you call downward compression?, as English is not my first language it can be a little confusing when it comes to technical terminology.

Big fan of paralell compression, especially in Live pa as it´s a lot more forgiving and it´s quite amasing how few professionals that know about it in my country..perhaps i should tell them or may be not /Toby
Old 27th August 2008
  #37
Audio Alchemist
 
Lagerfeldt's Avatar
Downward compression is simply regular compression.
Old 27th August 2008
  #38
Old 29th August 2008
  #39
Here for the gear
 

Quote:
Originally Posted by Lupo View Post
The problem is not that you've found a way to calculate equivalent values for parallel and regular compression. That's cool. It's the conclusions that are misleading.
Hi Andreas, I don't know why you see my conclusions misleading or problematic. I wanted to sound as neutral as possible. You probably expected a classic 'hooray-parallel-compression' article but it didn't happen. I can understand that. There are hundreds of them out there and basically they keep repeating the same things about extra transparency, super punch, 100% positive stuff. All their arguments sound very believable, I agree. But the truth is these articles consider the one side only - parallel compression on its on, they never compare it to downward compression. And if you do you can't help to say that actually, the king is naked.
BTW On your picture it seems you didn't take in account Make Up gain for Downward compressor. If you did these two graphs would look nearly identical.

Quote:
Originally Posted by Lupo View Post
Another point: Distortions may come from shaping the compressor response, but there is also inherent distortions in the actual units being used. A parallel comp mixed in at -12dB will also have 12 dB less of that distortion/noise. So parallel scheme does indeed lower some of the distortion. And lastly, the "hybrid curve" is not a mix of logarithm and linear, it's a mix of the compressor transfer curve and linear.
That is what I'm talking about. And why you don't consider what is happening in an equivalent downward compressor? Why not to mention that it's distortions are going to be lower just because it doesn't have work so hard? Is it my turn to call your conclusions misleading? :-)

Kind Regards,
Vitaly.
Old 29th August 2008
  #40
Here for the gear
 

kjg Have you considered buying that book rather then sharing the link for free reading? If you have some respect for the author, please delete it.

Quote:
Originally Posted by Paul Frindle View Post
I think your work is admirable and it explains the issues to people well.
Thank you, Paul.
Quote:
Originally Posted by Paul Frindle View Post
When you compress a track and mix it with the original what you are really doing quite simply is allowing the compressed path to bring up the levels in the quieter bits when the direct path is not loud enough.

But most importantly - you are allowing the direct (uncompressed) path to dominate for loud passages and events.

So this means that for loud punchy events you are mostly listening to the original signal - and filling in the 'gaps' with the compressed stuff which is less prominent and normally not a major signature in the character of rock tracks etc.
Confirmed. Here is the contribution of a compressor working in the parallel scheme with Thresh= -20dB, Ratio=2:1 and Make Up=0dB
@-20dB and below = 100%
@-15dB = 42%
@-10dB ~ 18%
@-5dB ~ 7.5%
@ 0dB ~ 3%
Thus, what we have is some kind of 'dynamic transparency'. The lower the signal, the dirtier it becomes. Or, the higher the signal, the cleaner it gets. Lets just remember one thing. The gain reduction of the compressor is circa -10dB. The gain reduction of the equivalent standard compressor (Thresh= -20dB, Ratio=1.22:1 and Make Up=6.02dB) is circa -3dB. Which makes me think that the standard compressor will have less distortion. Then, can we say for sure that parallel compression is clearer then downward one? I don't think so. What we can say though is that for loud events the parallel compression will sound clearer but for quiet parts the winner is downward compression. Again, we have a compromise. Again, we have to admit that by doing things differently we can win something, but we are going to lose as well.
Quote:
Originally Posted by Paul Frindle View Post
1. The contribution of the compressor to the loud and transient events is much less - so the timing characteristics of the compressor are much less of an issue.

2. The compressed path can have a wider range of timing settings without ruining the impact of the loud events - so you can do slower or faster settings without ruining the impression of the whole thing.

Therefore a far less 'capable' compressor can be used - simply because it's major contribution to the sound mostly doesn't happen during the impact-defining events. :-)
Not confirmed. You see, this is one of the issues when our logic goes against facts. In reality, the actual Attack and Release times of Parallel compression are in strict correlation with Attack and Release times of a compressor used in it. And they don't depend on the input signal level. I've done lots of tests and they prove the opposite. But I could make a mistake somewhere and if you can recommend some test proving your point I would appreciate it.

Enough talking. Lets have some fun, shell we? Here are four takes of the same part of a song. I admit three of them are compressed differently (one is the original) but amount of compression is the same. You say 'just listen'. Alright, lets have a listen. Could you tell which take has parallel compression?

Take1
Take2
Take3
Take4

Thanks,
Vitaly.
Old 29th August 2008
  #41
Lives for gear
 
Nordenstam's Avatar
 

Verified Member
Quote:
Originally Posted by Vitaly View Post
You probably expected a classic 'hooray-parallel-compression' article but it didn't happen. I can understand that. There are hundreds of them out there and basically they keep repeating the same things about extra transparency, super punch, 100% positive stuff. All their arguments sound very believable, I agree. But the truth is these articles consider the one side only - parallel compression on its on, they never compare it to downward compression. And if you do you can't help to say that actually, the king is naked.
My problem with the article is that it carries the same conclusion as your snippet above - "the king is naked". He's not. I don't have any particular for or against parallel, it's just another tool among with the screw drivers, hammer and angle grinder. What I don't like is the claim that they can give identical results. They don't.

If they where identical, they should null to a three digit figure, not to -40! -40 is nothing. That's about the same as 7 bits of performance.


Quote:
Originally Posted by Vitaly View Post
BTW On your picture it seems you didn't take in account Make Up gain for Downward compressor. If you did these two graphs would look nearly identical.
That is correct. I didn't make up gain, cause I don't know what make up gain to use. Since the transfer curves are not identical, I can either choose to match the level of the body of the sound, or the peaks. But I can not make up gain so that both body and peaks match - since the compression results are not identical..

The settings was threshold at -6dB and 2:1 ratio for parallel and 1.3:1 for regular comp, as suggested by your calculator. It also claims that 6.02dB make up gain should do it, but that would double the level of the signal, creating a curve like this:

Which is clearly way off. There's something for you to look into, why the calculator suggests something like that.


Adding 0.5dB make up gain looks like this:

Notice that the excursions both above and below the center line is bigger than in the parallel below.

Peak normalizing to 0dB (so the peak is the same as original) adds 1.1dB and looks like this:


None of those are the same as the transfer curve of the parallel in the original picture, included for clarity below:




<blatant rethoric> - If parallel is just a strange cloak for regular compression, how come it's not possible to create identical transfer curves using the different tools?

Perhaps it is because those things are not the same?


Quote:
That is what I'm talking about. And why you don't consider what is happening in an equivalent downward compressor? Why not to mention that it's distortions are going to be lower just because it doesn't have work so hard? Is it my turn to call your conclusions misleading? :-)
What I wrote was: "Distortions may come from shaping the compressor response, but there is also inherent distortions in the actual units being used." The distortions arrising from the transfer curve of the compressor is but one side of the coin. The units being used will have inherent distortions too. If you have a 1:1 ratio (no compression) there will still be distortions (signal independent noise plus harmonic and nonharmonic signal dependent errors). Those distortions will be lower in a parallel scheme.


Best regards,

Andreas Nordenstam
Old 29th August 2008
  #42
Here for the gear
 

Quote:
Originally Posted by Lupo View Post
My problem with the article is that it carries the same conclusion as your snippet above - "the king is naked". He's not. I don't have any particular for or against parallel, it's just another tool among with the screw drivers, hammer and angle grinder.
This is exactly what I'm trying to say! It's just another tool! It is not some suppa-duppa method suitable to improve everything around - drums, vocals, mixes, you name it - as it is presented in lots of pages on web. It is just another tool with it's own pros and cons. That's right.

Quote:
Originally Posted by Lupo View Post
What I don't like is the claim that they can give identical results. They don't.
Who did this claim? Me? No, I didn't! I always use the word 'nearly'.
Quote:
Originally Posted by Lupo View Post
Which is clearly way off. There's something for you to look into, why the calculator suggests something like that.
Because you use 0dB gain for the original signal and 0dB gain for the compressed signal in parallel compression. These settings will give you +6.02dB Make Up gain in downward compression. There is no mistake here. And yet, your picture of parallel compression doesn't show this, which means the actual gain of the original and compressed signals were ~-6dB when you did it. Check your settings.
Anyway, the compression curves won't be identical (and I've described why) they are very close. Close enough to be mistaken in a blind test.

Kind Regards,
Vitaly.
Old 29th August 2008
  #43
kjg
Lives for gear
 

Quote:
Originally Posted by Vitaly View Post
kjg Have you considered buying that book rather then sharing the link for free reading? If you have some respect for the author, please delete it.
If the author or publisher would have issues with the book being partly available for preview, I'm sure he could inform the retailer of that. I'm convinced that a preview snippet of a book might persuade people to actually purchase. That most likely is why it is there in the first place, and I don't think the author or publisher would mind the preview function of the store being used to point a newbie to a simple definition, meanwhile generating only more exposure for the book and its quality.

I simply wanted to demonstrate to mattssons that sometimes google turns up valuable answers in seconds, even to a question so basic (but fundamental), that he was even wondering if it might be silly to ask.

It is sad to see that this thread has come to you:
a) trying to portray me as a copyright infringer, while you know damn well I am only pointing out to a newbie how easy it is to use google AND exposing him to a great resource in the process.
b) trying to teach Paul Frindle(!) even something about parallel compression, a subject you yourself obviously haven't come to fully understand.

While the only thing you have to do is:
c) take the compliments for your work (and you've gotten quite some - more than it deserves actually)
d) admit that you've jumped to conclusions.

I don't understand how guys like Lupo still have the energy to reason with you. It has been spelled out in so many ways to you now, by so many knowledgeable people. You honestly still don't get it?? Do you still believe the conclusion of your "article" is not misleading and problematic??

I think you are just being stubborn and find it hard to admit you have jumped to conclusions in your enthusiasm. That's ok... It's common in young ambitious men. So stop posting in this thread, then. Maybe in time you'll understand there is no shame in just admitting you were only partly right. It just makes you a bigger man. But resorting to polemics and even using a disrespectful tone to one of the most knowledgeable and respected members of this forums? I have no respect for that.

So what is it? Are you stubborn and proud or just a bit slow?

I own both the first and second edition of the book by the way, thank you very much.
Old 29th August 2008
  #44
Audio Alchemist
 
Lagerfeldt's Avatar
Quote:
Originally Posted by kjg View Post
b) trying to teach Paul Frindle(!) even something about parallel compression, a subject you yourself obviously haven't come to fully understand.
I don't think Vitaly knows who Paul Frindle is. At least that would explain something.
Old 29th August 2008
  #45
Lives for gear
 
Bob Yordan's Avatar
Quote:
Originally Posted by Paul Frindle View Post
I think your work is admirable and it explains the issues to people well. But I think in reality the concept is far simpler than this discussion seems to be indicating? And no - it cannot ever be the same thing as pushing the same track through a single compressed path - unless the compressor used is devised deliberately to mimic parallel compression. I don't know of any off-hand.

When you compress a track and mix it with the original what you are really doing quite simply is allowing the compressed path to bring up the levels in the quieter bits when the direct path is not loud enough.

But most importantly - you are allowing the direct (uncompressed) path to dominate for loud passages and events.

So this means that for loud punchy events you are mostly listening to the original signal - and filling in the 'gaps' with the compressed stuff which is less prominent and normally not a major signature in the character of rock tracks etc.

Quite apart from the input/output gain characteristics of this - which you cover well - the major reasons this will sound different to setting the same input/output curve for a single comp over the whole signal are:

1. The contribution of the compressor to the loud and transient events is much less - so the timing characteristics of the compressor are much less of an issue.

2. The compressed path can have a wider range of timing settings without ruining the impact of the loud events - so you can do slower or faster settings without ruining the impression of the whole thing.

3. Therefore a far less 'capable' compressor can be used - simply because it's major contribution to the sound mostly doesn't happen during the impact-defining events. :-)


The biggest downside to compressing this way on a master buss is that it does not control the peak levels - and this is a big issue in a world where the 'red light' has become everything..
Yup.

Good post.

Old 29th August 2008
  #46
kjg
Lives for gear
 

Quote:
Originally Posted by Bob Yordan View Post
Yup.

Good post.

Easy now, Bob! We don't know for sure if this is a good post since Vitaly hasn't confirmed this information yet! You should read his posts much more closely. tutt

You think it is good... But the "facts" prove our logic wrong, as Vitaly has explained before. Did you even read the thread?

Tssss...

Last edited by kjg; 29th August 2008 at 01:59 PM.. Reason: !!! unconfirmed information warning !!!
Old 29th August 2008
  #47
Lives for gear
 
Bob Yordan's Avatar
Quote:
Originally Posted by kjg View Post
Easy now, Bob! We don't know for sure if this is a good post since Vitaly hasn't confirmed this information yet! You should read his posts much more closely. tutt

You think it is good... But the "facts" prove our logic wrong, as Vitaly has explained before. Did you even read the thread?

Tssss...
I just liked what Paul wrote.

No matter what proves wrong or not.

When I develop my plugs I use all kind of parallell ways of processing signals. Not just blending signal paths but even calculating merging gain levels controlling the dynamics processing. Mostly this occures in multiple stages. Including all kind of different signal processing.

Old 29th August 2008
  #48
kjg
Lives for gear
 

Quote:
Originally Posted by Bob Yordan View Post
When I develop my plugs I use all kind of parallell ways of processing signals. Not just blending signal paths but even calculating merging gain levels controlling the dynamics processing. Mostly this occures in multiple stages. Including all kind of different signal processing.

I know what you mean, I do sound mangling in similar ways. I use tools like Max/MSP, SuperCollider, Csound.

Do you use any higher level languages or do you develop in C++ straight?
Old 29th August 2008
  #49
Lives for gear
 
Bob Yordan's Avatar
Quote:
Originally Posted by kjg View Post
I know what you mean, I do sound mangling in similar ways. I use tools like Max/MSP, SuperCollider, Csound.

Do you use any higher level languages or do you develop in C++ straight?
Actually I use Synth Edit, with my own specially made modules (C++).

Most of my plugs loads 50-70% of the cpu in eg a 3.2GHz P4 Intel PC.

One of my friends that tested a plug in his Mac Pro, discovered that it
load 60% of his cpu also. heh

The bouncing of a 4 minute long mix takes approx 3 minutes in Wavelab.

Old 29th August 2008
  #50
Gear Addict
 
hackenslash's Avatar
 

Quote:
Originally Posted by kjg View Post
Easy now, Bob! We don't know for sure if this is a good post since Vitaly hasn't confirmed this information yet! You should read his posts much more closely. tutt

You think it is good... But the "facts" prove our logic wrong, as Vitaly has explained before. Did you even read the thread?

Tssss...

Ma! Put some popcorn on! heh
Old 29th August 2008
  #51
Lives for gear
 
Bob Yordan's Avatar
Quote:
Originally Posted by hackenslash View Post
Ma! Put some popcorn on! heh
Old 29th August 2008
  #52
Lives for gear
 
Nordenstam's Avatar
 

Verified Member
Quote:
Originally Posted by Vitaly View Post
This is exactly what I'm trying to say! It's just another tool! It is not some suppa-duppa method suitable to improve everything around - drums, vocals, mixes, you name it - as it is presented in lots of pages on web. It is just another tool with it's own pros and cons. That's right.
Can you please link to some of those glorifying articles? The one I've seen have pointed out the differences and what can be achieved with the lesser known way of doing compression as compared to the regular one. No one said it would do miracles. Can understand your disappointment if you expected it to do the dishes too.. :D

Quote:
Who did this claim? Me? No, I didn't! I always use the word 'nearly'.
As I already noted, it's not expressively written anywhere in your text. But that IS the total sum of what you write. Quotes from your two last posts to me: "And if you do you can't help to say that actually, the king is naked." "Anyway, the compression curves won't be identical (and I've described why) they are very close. Close enough to be mistaken in a blind test." Your text does the same arguments as well. Anyone reading that paper of yours (and your arguments here) will probably end up with the impression that the difference is so small as to be undetectable. What is there for you or anyone to gain from that? Do you plot on a monopoly on parallel?

Quote:
Because you use 0dB gain for the original signal and 0dB gain for the compressed signal in parallel compression. These settings will give you +6.02dB Make Up gain in downward compression. There is no mistake here. And yet, your picture of parallel compression doesn't show this, which means the actual gain of the original and compressed signals were ~-6dB when you did it. Check your settings.
Checked my settings. Your calculator is right. My bad. The parallel picture was a 50/50 mix, not 100/100. With a 100/100 mix the downward had to be raised 6.02dB to compensate indeed. Though, the end result was exactly the same as in doing a 50/50 and not using make up gain. The curves where identical to the first transfer curve picture I posted.

It doesn't seem to me that you even bothered to try this yourself..?? Can you procure forth any such real curves for parallel and regular compression that shows them to be similar? The paint drawing in your paper doesn't quite do it in the convincing department.

Quote:
Anyway, the compression curves won't be identical (and I've described why) they are very close. Close enough to be mistaken in a blind test.
They are close indeed, both are compressing after all. Though that's a far stretch from claiming they can not be discerned in a blind test! The same argument can be used to claim that two different brands of compressors are the same. They'll probably null to lower than -40 if the settings can be made to be the same. The problem is all yours if you do not hear a difference between various pieces of gear and various processing options. That doesn't mean that the rest of the world doesn't. A null to somewhere below -90 would make it more interesting to do A/B/X tests. If they null to below at least -100, I personally would strive for -120 or so, they can be said to give the same real world result.


As for the four .wav's you posted, found it hard to get a clear grasp on the compression action on the electronic sound. It's easier to hear compression for me on real instruments. It would at least make it easier if you selected a part with both breakdown and chorus, that's where the differences make themselves best heard. It seemed to me to be a consequent difference between two pairs of files, but haven't done seamless A/B/C/D switching so I can't really claim which is which. If you want me to test for difference - name two of the files and I'll put them through the paces in the PCABX software. Four are too many for an A/B test.


Quote:
Originally Posted by kjg View Post
I don't understand how guys like Lupo still have the energy to reason with you.
Just being stubborn. Hehe.. Totally off topic, just get a snicker out of the energy comment. The vigour level is (f)actually almost ridiculous! Been supporting the music and mastering by doing part time bicycle couriering for nearly six years now. And it's true what they say, that exercise gives more energy..


@ Bob Yordan: Bet you'll still use 70% CPU even if you get quantum computing in a cloud! heh


Regards,

Andreas Nordenstam
Old 29th August 2008
  #53
Lives for gear
 
Bob Yordan's Avatar
Sorry for being off topic.

Quote:
Originally Posted by Lupo View Post

@ Bob Yordan: Bet you'll still use 70% CPU even if you get quantum computing in a cloud! heh


Regards,

Andreas Nordenstam
Well I had to remove a clipper algoritm a while back because the CPU
load got up to 95%. heh And the mix kind of started to stu-ut-ut-tter when played back in WL6 and the PC froze with a terrible noise emitting from the sound card. That would indeed pop the clouds from the sky (no more rainy days). It was kind of cool to see his face mimic, because he bragged
just before he started to test the plug how much extra cpu power he had in his new DAW and that my plug would not cause any problems in that area.

Old 31st August 2008
  #54
Lives for gear
 

In rock/pop etc, the advantage of blending upward compression tends to be decimated when the ol' loudness limiter/clipper destroys those lovely little transient peaks you worked so hard to retain. Am I the only one here to experience this?

Still, I think Parallel compression CAN be made to sound quite different to downward compression by the way you eq the returns. Now that presents a whole new "phase" ball game, just ask Michael Brauer!
Old 7th September 2008
  #55
Here for the gear
 

kjg, I apologize about the book. You are right, it is a preview only. I guess the way it's done (badly scanned) is what confused me. I use to find pdf files as a preview. My bad.

Lagerfeldt, I know who Paul Frindle is and I do respect his works and ideas. And I can't see how I'm trying to teach him. Because I couldn't confirm one of his points and I dare have my own point of view? After all, that is how people talk – you've got an idea, I've got an idea, lets have a look. I'm always opened for a constructive dialog and that is way I wrote 'But I could make a mistake somewhere and if you can recommend some test proving your point I would appreciate it.'

Dear colleagues, I can understand your feelings. You've read my work and you don't agree. Great! Can't we have a different opinion on the matter? Or I'm telling you to stop using parallel compression as you are telling me to stop writing here, on my own topic? No, I'm not. All what I want is better understanding of this method as I believe there is a big difference between what we think and what is. And yes, this is my personal opinion only. As personal as yours though.

My best wishes to all of you,
Vitaly.
Old 7th September 2008
  #56
Lives for gear
 
Alexey Lukin's Avatar
 

Verified Member
Vitaly, you should take a look at this manual, page 15. It clearly shows that the transfer curve of a parallel compression is very different from the transfer curve of a downward compression: it is S-shaped. If your calculator fails to demonstrate this, maybe your scales or parameters are not tuned properly?
Old 8th September 2008
  #57
Here for the gear
 

I see. Aleksey Lukin, it is simply the matter of gain makeup. If parallel compression Make Up is zero (for 50/50 mix, thr= -50dB, ratio= 2.5:1, dry= -11.75dB and comp= -11.75dB), then its transfer curve will look as it's presented on pic.7(dashed curve) in my article. If we increase dry and comp levels so that a full level signal output (0dB) will match a full level signal input (0dB), then the transfer curve will look like it's shown on that manual. That's it. Of course, the way the transfer curve looks is also the matter of the scale we use – logarithmic or linear.

The calculator doesn't calculate the shape of the transfer curve. It's simply trying to match parallel and downward transfer curves at two levels – Threshold and Gain Reduction.

Thanks for your input.
Vitaly.
Old 8th September 2008
  #58
Lives for gear
 
Storyville's Avatar
This Calculator is Very Cool

Quote:
Originally Posted by Vitaly View Post
bmanic, where I called it 'near identical'? I said: 'There is a very small difference. Whether it is worth setting up the parallel scheme is totally up to you.'

BTW I'm happy with my Genelec. Thanks.

Guys, apparently you've got it wrong. Nobody is going to take Parallel compression from you My goal was to share the results I've come to as I thought it could be interesting to someone else. There is no need for anger here.

Kind Regards,
Vitaly.

I A/B a Parallel setup with a Downward setup on a drum track using Vitaly's calculator. The sense of transient changed pretty dramatically in my ear. I also found the makeup gain calculation to be way off. Maybe I misunderstood the use of the calculator. However, I re-calibrated the makeup gain subjectively (and then checked it against my peaks and rms) to see the difference. When matched in level, the downward comp had a much thicker sound. My theory on this is that many compressors cannot be set to a ratio of 1.0533, and when your threshold is -50db, 1.0533 and 1.05 have very noticeable differences.

My conclusion is this, I think that the research needs to be revisited, or I need a better explaination on using the calculator. BUT MORE IMPORTANTLY, the invention is often more important than the hypothesis. I think I will keep this calculator handy - my hyphothesis is that I can find a good setting using parallel comp, and if I want a thicker sound use the calculator to find an equivalent downward comp - I'm thinking a good tool for making a large bass sound....

But yeah, there's a bit of ego in that article you may want to consider revising. Then again there's a bit of ego in some of the responses. Thanks for the research, it's very interesting and this calculator is very cool.
Old 8th September 2008
  #59
Lives for gear
 
Storyville's Avatar
And for the 4 tracks

It's difficult to tell. My instinct tells me that Take 2 sounds like Parallel Compression, while 3 and 4 sound more like regular Compression. Take 4 seems to have the most body, a bit more apparent volume.

How'd I do?
Old 8th September 2008
  #60
Lives for gear
 
Alexey Lukin's Avatar
 

Verified Member
Quote:
Originally Posted by Vitaly View Post
it is simply the matter of gain makeup.
A simple gain makeup will not turn S-shaped curve into Г-shaped. This upper segment of S-shaped curve, which is (almost) parallel to the x=y line, makes the difference between parallel and downward compression. And it may or may not be a small difference, depending on the parameters.
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
firstclizzy / Rap + Hip Hop engineering and production
10
crabtwins / Mastering forum
2
Watersound / So much gear, so little time
3
Tim Farrant / So much gear, so little time
15

Forum Jump
Forum Jump