Does Parallel Comp. Really work?

[hduncan]

As I think about it, it just seems like you'd get the same sound by using just one track compressed with a lower ratio. Or does the "sound" come from the layering of a different attack/release between the compressed and umcompressed?

If not, all youre doing by squashing one and adding it to the uncompressed is shortening the dynamic range- which you don't need PARALLEL compression for.

What am I missing?

[Sensual Ears]

have you done it?

it works, i guess you'll just have to trust me!

[dviola]

Originally Posted by DarkEcho

the "sound" come[s] from the layering
What am I missing?

Nothing.

give it a try

tell us how it sounds!

[andychamp]

Originally Posted by DarkEcho

As I think about it, it just seems like you'd get the same sound by using just one track compressed with a lower ratio.(...)If not, all youre doing by squashing one and adding it to the uncompressed is shortening the dynamic range- which you don't need PARALLEL compression for.

as long as you keep the level of the squashed signal below or equal to the lowest uncompressed signal, you'll still have the original dynamic range.
But you're adding a compressed and often gated and severely EQ'd version of the same sound.
This will help you "anchor" a drum - or other - track in the mix without sacrifying it's dynamics.

[Nordenstam]

Originally Posted by DarkEcho

If not, all youre doing by squashing one and adding it to the uncompressed is shortening the dynamic range- which you don't need PARALLEL compression for.

Both aproaches shortens dynamic range. With parallel, this comes from pushing the bottom up and leaving the peaks intact - the opposite of normal compression.

[luctellier]

Short answer is: YES

[warhead]

Yes, think of it as "bottom up" as described by Lupo.

Parallel compression has no parallel...!

Sorry, bad pun.

War

[hduncan]

Thanks everyone.
I undserstand this better now with the bottom up analogy. If anyone is up to it, I'd like a more scientific explanation of the differences between normal and parallel compression involving amplitude maybe.

I'm going to do some tests and watch the effects on the waveform.

Also- how can I ensure that, after compression, the peaks are at the level they were originally, using make up gain in relation to my threshold/ratio settings?

[Mike Caffrey]

Originally Posted by DarkEcho

As I think about it, it just seems like you'd get the same sound by using just one track compressed with a lower ratio. Or does the "sound" come from the layering of a different attack/release between the compressed and umcompressed?

If not, all youre doing by squashing one and adding it to the uncompressed is shortening the dynamic range- which you don't need PARALLEL compression for.

What am I missing?

Technically, you don't need compression for shortening the dynamic range either.

Essentially your summary is right, but you won't achieve the exact same sound.

The compressors tonal attributes will be diluted.

The evenlop shaping will be diluted, and in the case of a fast attack setting, with have no audible effect at the attack stage because the uncompressed signal will mask the compressed signal.

As dynamic range increases the balance between your compressed and uncompressed signals will change because the compresses signal will be unable to increase like the uncompressed one does. At low levels you will hear more compressed signals and at high levels you'll hear more dry signal.

That's why I think the best approach in many cases is to compress the sum of the original and parallel signals.

[leaper]

Originally Posted by DarkEcho

Thanks everyone.<br />
I undserstand this better now with the bottom up analogy. If anyone is up to it, I'd like a more scientific explanation of the differences between normal and parallel compression involving amplitude maybe.<br />
<br />
I'm going to do some tests and watch the effects on the waveform.<br />
<br />
Also- how can I ensure that, after compression, the peaks are at the level they were originally, using make up gain in relation to my threshold/ratio settings?

<br />
<br />

No man... you're adding compressed signal to uncompressed signal. if it's not a rare compressor with a mix function, this means that both the compressed and uncompressed signals will be coming in on different faders.. you add the compressed so that it's lower in volume than the uncompressed, in that way the peaks of the uncompressed are preserved. That's why it's called "parallel". If it's a hardware box with a "mix" function ( or a plug in like Vintage warmer ) then both compressed and uncompressed signals are being summed.. same principle.

[Mike Caffrey]

Originally Posted by andychamp

as long as you keep the level of the squashed signal below or equal to the lowest uncompressed signal, you'll still have the original dynamic range.
But you're adding a compressed and often gated and severely EQ'd version of the same sound.
This will help you "anchor" a drum - or other - track in the mix without sacrifying it's dynamics.

Not true. You're narrowing the dynamic range by bringing up the bottom without capping off the top.

The umcompressed signal will have full dynamci range, but the compressed one will not, them sum will be in the middle.

The other way to think of it is how they mask each other.

Let's say your a a mid level dynamic range and you set the parallel signal so that when you alternate between the two they are identical in output.

At a quieter section, like the intro, the uncompresed signal may drop a lot, but the compression in the parallel side with keep the volume up and mask the uncompressed signal.

When you go to a louder section, the uncompressed signal will get louder, but the compressed signal will run into the limit imposed by the gain reduction. and the uncompressed signal will mask the compressed signal.


Actually the best way to look at it is that your maximum dynamic range is the noise floor to infiinity (or what your gear is capable of). When you add in the parallel compression, you'll be increasing the noise floor - even when there's no signal, there's noise for the compression to amplify. That's decreases the range bewttn the noise flloor and infinity/the max your gear is capable of.

[luctellier]

Mike couldn't have explain it better! thumbsup

[James Meeker]

It's not the right technique for every session. Just like every other method out there. This artistry is picking and choosing your approach.

Parallel compression is not a "make or break" kind of deal. You have to have good sounds, good mix, etc... going into it. The main benefit is added "thickness" of the sound. It can add weight.

Lots of things can add weight though, this is just one of them.

I find the greatest success with using a relatively slow attack (20-40ms) on the compressor, high ratios and long release (lets the peaks "punch through" on the transients). I also find that "the zone" for the parallel compression is about 12 db underneath the main drum buss.

If you smash it too much with too quick of a release you can sometimes add clutter to the percussion. If the release is too slow you can expect pumping and breathing as well.

If you haven't tried it with a slow attack you may wanna give that a whirl, just to see how it works for future use.

Hope this helps!

[Nordenstam]

Originally Posted by DarkEcho

If anyone is up to it, I'd like a more scientific explanation of the differences between normal and parallel compression involving amplitude maybe.

SIDE CHAIN COMPRESSION

[Vogon]

Originally Posted by Mike Caffrey

As dynamic range increases the balance between your compressed and uncompressed signals will change because the compresses signal will be unable to increase like the uncompressed one does. At low levels you will hear more compressed signals and at high levels you'll hear more dry signal.

That's why I think the best approach in many cases is to compress the sum of the original and parallel signals.

Yes, well put. That and the fact that the cymbals retain sizzle, and things don't duck and pop on extremely loud transients is why I've been using it recently.

[hduncan]

This goes into compression 101, but, When used to control level (IE Smash the transient peaks down so you can bring the entire level up), if you have a slow attack, aren't you letting through the part that you are trying to squash?

[zoink]

parallel isn't for everything but it is very effective for quiet-loud-quiet-loud songs.

[bgrotto]

Originally Posted by DarkEcho

the effects on the waveform.

?!?!?

Don't WATCH anything! Listen to what it's doing, and especially check out the spaces between transients (pumpy release times can add a lot of excitement to a drum track).

You can also use it on a mix; it's often used in jazz and classical styles (quick attack and release times with low ratios to bring up details).

One of my favorite applications is sending drums as well as bass to a shared parallel compression buss. It really glues the rhythm section together.

[hduncan]

I don't know how long some of you have been doing this, but I am a beginner and I still have to "teach" my ears. Since I learn well with visuals, I want to WATCH the effect compression has on a waveform, how it shapes it, and what it does to certain parts.

I see nothing wrong with watching what I hear to teach my ears...... Yet people tell me to listen-listen-listen all the time.. How will I even know what I am hearing until my ears learn?? Visuals will help me learn fast so that my brain can better interpret what my ears are hearing.

[bgrotto]

Originally Posted by DarkEcho

I don't know how long some of you have been doing this, but I am a beginner and I still have to "teach" my ears. Since I learn well with visuals, I want to WATCH the effect compression has on a waveform, how it shapes it, and what it does to certain parts.

I see nothing wrong with watching what I hear to teach my ears...... Yet people tell me to listen-listen-listen all the time.. How will I even know what I am hearing until my ears learn?? Visuals will help me learn fast so that my brain can better interpret what my ears are hearing.

Fair, however the waveforms displayed in your DAW are merely approximations of something far too complex to relay visually with any accuracy. If you apply some parallel compression to a track, then bounce it, it won't look that much different than mix compression or perhaps limiting. It may not even look that much different than an unprocessed mix. The point is, waveforms can be great for finding transients to edit drums and such, but they're good for little else. If you're looking to pursue engineering as a career, try to get comfortable with learning with your EARS as well as your eyes. You'll learn much more, much more quickly. This is particularly true if you find yourself working as an intern (which I HIGHLY recommend). Most engineers won't have the time (and many won't have the inclination) to explain what they're doing at each step of the process. If you're competent in your hearing, you'll learn a great deal without having to ask questions that will irritate the engineer.

[Daedalus77]

I think Mike Caffrey in fact talked in detail in a recent TapeOp article about his tactic of compressing the sum of the original and parallel signals. I think it was a couple issues back...you should check it out.

[pootkao]

I disagree completely.
I've spent the last couple years trying to "unlearn" and STOP watching.
I turn the monitor off as often as possible when mixing, particularly for clients, as everyone just sits there googly eyed watching the monitor and not listening.

I used to give the same argument you just gave (below) but have now realized its complete bunk. We listen to music. We don't watch it. Yes, its a handy tool to be able to edit etc. But mixing should be done by ear, even if you spend 10 years screwing it up as your ears learn.

My 2 cents.

Originally Posted by DarkEcho

I don't know how long some of you have been doing this, but I am a beginner and I still have to "teach" my ears. Since I learn well with visuals, I want to WATCH the effect compression has on a waveform, how it shapes it, and what it does to certain parts.

I see nothing wrong with watching what I hear to teach my ears...... Yet people tell me to listen-listen-listen all the time.. How will I even know what I am hearing until my ears learn?? Visuals will help me learn fast so that my brain can better interpret what my ears are hearing.

[George Necola]

if you are parallelcompressin ITB you have to put the same compressorplugin at the "other bus" and bypass it or you will have time/phase/latency issues.

for example you have

drumbus A (pluginchain: eq-plugin=>comp XY=>transientsomethingplugin)

and

Parallelcomp NY ultratrick A (pluginchain: eq-plugin=> comp ZW)

you need to add the comp XY to the parralelcomp NY ultratrick A bus and bypass it.

sorry for my bad expl... it's early here.. brain still filing with blood

[Vogon]

Originally Posted by George Necola

if you are parallelcompressin ITB you have to put the same compressorplugin at the "other bus" and bypass it or you will have time/phase/latency issues.

Yes, this'll depend on your DAW's delay-compensation scheme though. Ie. In logic (7+), it can be configured to compensate automatically on bus's.

[kats]

Don't bet the farm on ADC... take the extra time and play it safe.

[George Necola]

Originally Posted by Vogon

Yes, this'll depend on your DAW's delay-compensation scheme though. Ie. In logic (7+), it can be configured to compensate automatically on bus's.

doesn't work properly in Cubase 4 and Samplitude 9...

[hduncan]

Here is a quickie semi-related.

What is the difference between Peak and RMS setting on a compressor? which should be used when?

[dubrichie]

Originally Posted by DarkEcho

Thanks everyone.
I undserstand this better now with the bottom up analogy. If anyone is up to it, I'd like a more scientific explanation of the differences between normal and parallel compression involving amplitude maybe.

I'm going to do some tests and watch the effects on the waveform.

Also- how can I ensure that, after compression, the peaks are at the level they were originally, using make up gain in relation to my threshold/ratio settings?

the idea is to use a fast attack on your parallel compressor. you want to "squash" the transient peak out of the sound so that you're just left with the sustain or "body" whch you can add gain to.

when you mix this back into the original, uncompressed sound you will have the transient peak of the original intact but the sustain is supplemented by the that of the compressed, amplified "parallel" signal.

so, you still get the dynamic "punch" of the original but also more "body".

if done correctly the peaks should be at the same level as they were before.

capice?

and don't LOOK for the difference in the DRAWINGS of the waveforms!

LISTEN for it in the SOUND!

good luck.

[dubrichie]

Originally Posted by DarkEcho

Here is a quickie semi-related.

What is the difference between Peak and RMS setting on a compressor? which should be used when?

are you talking about the difference when concerned with the action of the compressor on the audio, as opposed to the behaviour of just the meter?

if so then the PEAK setting will mean that the compressor will respond to peak signal levels and the RMS setting means that it will respond to the RMS averaged level of the signal.

this will affect the attack and release settings and behaviours of the compressor... i think... sort of? i'm not sure if i explained that satisfactorily, apologies.

[Vogon]

Originally Posted by kats

Don't bet the farm on ADC... take the extra time and play it safe.

Yes, I agree. But I did a slew of tests at one point, and it appears to work in logic, once set-up properly... I tried the manual tweaks, including sample-delay/shifting, and logic adc seemed right on the money - in these specific cases.
It's just a shame apple appears to be steering some other aspects of logic more towards the hobbyist/dj market, with some of it's changes in V8.