If I understand the concept of parallel compression correctly, it is basically mixing a set of a clean track in alonside that same track compressed, mixing to taste.

IF that is so, doesn't that cuase phase problems?

Can someone perhaps explain or detail it out for me a bit more?

Thanks,

-a

DIRTY HALO www.dirtyhalo.com

Only if there's delay caused by the compression, either from latency caused by a plugin compressor, or by a trip out the D/A converter, to the hardware comp and back in the A/D.

Often, in a DAW, people will open the same comp. plugin on two tracks and set one to bypass.

But be careful; some plugins don't cause latency in bypass, and in those cases you have to set something to 'no compression' like the ratio or the threshold.

When running parallel compression, one channel dry and the other compressed, wouldn't the two still have some kind pf phasing problems?

-a

no.

I think technically it might - depending on how u did it ... and even then each bit of outborard has " some" latency ... how relevant that microscopic latency is another thing.

Would you use a send ... or assign a bus ?? ...

phasing only happens when you have some sort of delay (very small) on one signal and not the other...., and you mix the two signals.....

If no delay, no phasing.

I may not be calling it the right term, but if you stack identical waves in sync, there is wave amplitude problem (oh man, my engineering theory is way back in the brain). ANYWAY, on parallet compression, is there any of this, any problems or do you find it works great... ideally I'd love answers from a few people who do parallel compression.

Thanks!

-a

Then it gets LOUDER.

If in phase, and no delay

I've been using parallel compression for quite a few years with no problems on an analog board. Any delays are so short that they would only affect extremely high frequencies effectively above the operating range of our systems and hearing.
I use it as I would an effects send/return and commonly have a Manley Vari Mu parallel to drums and bass, sometimes a Compellor on acoustic guitars etc. and if I've got a rough and tumble tune on my hands a couple of Sta Levels parallel to the elec guitars.
As to adding "identical" waves.....think about it. Whats coming back from the compressors is not identical to the original waves. That's the whole point.

Someone here on Gearslutz recently suggested that phase problems are possible when using parallel compression on submixes/busses and advised to use it on individual tracks instead.

Any truth to this???

I use it on mutiple instrument groups, both mono and stereo, all the time with zero problems. Obviously, on stereo groups you've got to pay attention to the relative panning of the source and parallel return.
Once you get comfortable with parallel compression you can start using eq on the parallel send or return to really help with certain problems. A parallel with a multiband compressor is also very useful at times.
A more involved parallel scheme is to take a source and spread it over several faders and comp/eq each independently for specific frequency ranges and combine at the board. This was detailed in an article in Tape Op about a year ago.
Give parallel processing a try. I find it one of the best "tricks" I've picked up in my thirty five years in the business.

Careful...

once you start throwing EQ into the equation... THEN there can be phase problems... still no time delay, but eqs mess with your phase. if someone makes an analog eq that doesn't mess with the phase, i'll give them france.

Example:

4 multitrack returns bussed to group out A, as well as the main stereo buss. Group out A is now send to a comp. Comp return now brought back to a fader and mixed to the main stereo buss.

this is cool. this is sweet.

make sure that you patch everything right... and for henry ford's sake, don't buss the comp return to Group A. That kinda sucks... AAAHHAHAHAHAHA!!!

now, if you throw some eq onto the comp return, and mess with it... the phase of the comp return gets messed with. not 'technically' a good idea.

that can be a bad thing. it can be a good thing... hell play with it, if it sounds good, feck it... we're in ireland.

Is everyone confused or is it just me who'se confused?
The application of parallel compression is adding together the characteristics of two different compressors; so what you end up with is still just a compressor, but possibly with a more complex set of attack, release and ratio characteristics.... but it's still a compressor..... so there was something wrong with the original one you bought?
I suppose that if you have a favourite compressor that does a particular job very well, but not much else, then it would be reasonable to parallel it with a radically different compressor... but thinking deeper about it, I'm not so sure.... in fact I think it's a waste of time.... Hmmm.

As for introducing EQ as well...... apart from my opinion that EQ should be treated as a last resort, and most should be banned and covered in health warning labels, all musical EQ works by shifting phase (that's true phase shifting, not simple time delay) so mixing an EQ channel and a non-EQ channel will merely give slightly modified EQ curves, and certainly could be arranged to give a different musical spectrum according to the dynamics of the program...... in fact, it's a much more sensible idea than just paralleling compressors; but there certainly won't be a commercial product that does it! Hmmmmm..............

Ted Fletcher
www.tedfletcheraudiogear.com

Ted

I think whats being described is sending a lightly, or uncompressed stem to a compressor, compressing very hard (perhaps for example to emphasize transients on a drum bus) and them bringing the signal back to a pair of faders and mixing that squashed signal to taste with the original.

Thats the way Im interpreting the question anyway

That's what I thought parallel compression was. Sending some of the signal to outboard equipment and mixing with the original signal.
I have tried to send by ADAT to my KSP8 compressor algorithms and see phase everytime. There is no conversion going on, since I'm sending digitally, but just the processing through the KSP8 must add a slight delay causing phase problems when mixed with the original signal.
My solution (until I find out if this can be compensated properly in Cubase SX3)--just send the whole signal and using the Send % in the KSP8. Thus, maintaining some of the original signal. I guess that is not parallel compression then?

SX3 features external delay compensation. Hit f4 and go to the appropriate tab.

You will be able to add your KSP8 to the list of external devices, and then it will allow you to ping the KSP8 - this will measure the latency and record it so that the correct amount of compensation happens.

bear in mind that different algorithms in the KSP8 will likely induce differing amounts of latency and as such you'll need to re-ping often and between projects.

As far as I understand it, there is an uncompressed signal (A) and a parallel heavily compressed signal (B) mixed together in order not to affect the transients in the uncompressed signal (A) but to bring up all the low level mudd in the compressed signal (B). The "overcompressed" signal (B) will also alter the relationship between direct sound and room sound, by bringing up the room sound more. What you end up with is a mix that still sounds open but also has some urgency and energy because all the low levels are brought up in the compressed signal (B). I set the compressor to fast attack and fast release, so it pumps quite a bit. So far I haven't found a better compressor than the Chandler TG-1 for this. The Ridge Farm Boiler comp comes close, so does the AlSmart C2 in "Crush" mode, but the TG-1 still adds something the others don't. In fact the RNC does a great job on this too.

Problems with phase can show up when working in the digital domain, because signal (B) might be delayed by latency of the plug in or the D/A - A/D converters used to send and return from the anlog compressor. The solution is to delay signal (A) and ALL other tracks in the mix by the same amount. To find out the dealy time, switch your (analog) compressor to bypass and reverse polarity on the "to-be-compressed" channels. It's easier to use only one side of each signal for this test (i.e. only the left side) or just the snare. Both signals should be at the exact same level. Then bring up the delay in the channel of the "uncompressed" signal slowly until they cancel out, or at least almost cancel out, there will be a slight difference because (B) is going through an analog chain. Work in the sample range, it's about 20 to 90 samples at 48K, depending on the latency of the converters in my system, ms will be to course. The resulting delay will be the value to be used on all other channels in your mix setup.

In Nuendo (SX3) you can just "ping" the analog chain and it will give you the proper delaysetting for all other tracks and apply them if you so desire.

I think you mean phase cancellation, no phasing problem...Two identical waves playing together cancel each other resulting in a dramatic loss of dynamic and amplitude.

Merging the original track with the return from an outboard compressor gave me phasing despite of time aligment.
Some of the advanced GS pals explained to me then that the coloration of the circuits would inevitably lead to phasing when mixed with the original, and that this could only be omitted by sending the clean track through the same path, only without processing. Then mix the two.

In other words, you can´t have it without phasing unless the original track was recorded through the same circuits.

Sounded plausible to me.

Ruphus

oops, streetbeats, you beat me to it, took me a while to type the reply

My expereince differs on that. There will be some "sound" differences, especially in the low end, because one set of channels is going through an D/A - A/D conversion and some analog gear, but you should be able to align phase with the use of very short delays.

Rufus, it depends how you were sending and returning. Time aligning a copy of the track doesn't allow for the latency of the 2 conversions that take place going to analog outboard and returning. If the daw has auto delay compensation and you've got it set up right it works fine. Except for sends. It's a lot easier on an anlog console.

Thanks for your input, guys.

So, you are saying that the characteristical differences of the two tracks won´t cause phasing ( time alignment provided )?

Ruphus

Even if there's an auto delay in the DAW, visually aligning the parallel compressed track with the uncompressed track would be advisable. This means going into a somewhat extreme magnification and choosing a zero crossing or very clear peak and lining up both tracks to that exact moment. I personally wouldn't trust an auto delay for that.

Parallel compression is easier to deal with in the analog world, because you don't have to make sure your tracks are perfectly lined up every single time you make a pass. Which is the problem with plugin parallel compression: you really must record the pass, then line it up manually, and then listen back.

This business about the parallel tracks needing to be sent through the same circuits as the original is just not correct in practice. It may seem plausible in theory, but in reality it has no bearing on the matter whatsoever.

The bottom line is, when done properly, there are no phasing artifacts using parallel compression. But it really is cumbersome to use parallel compression in a plugin environment, or a mixed digital/analog environment where one of the paralleled tracks has to travel through converters and the other doesn't.

Thanks Streetbeats. I'm new to Cubase SX3 and still figuring out this KSP8. Up to now I was had ADAT's through KSP8 as input and output busses rather than using the external effects option. I tried what you said and it worked well.
I pinged for the delay, it detected--and voila--no delay.
I don't see this as parallel compression, though, since it seems to be sending the whole signal to the KSP8 and back (not just part of the signal like I had when I was using the Send channels in Cubase SX3).

is anyone tracking with parallel compression?
if it is something you know know you will be using.
use a mult from your patchbay and go to the comp. them record both as separate tracks.
i have not tried parallel compression but i plan to.

what is pinged for the delay?

Heres what I understand on this topic:

Youll always get phase shifts at different freqs when you go through any type of processing gear. This can range from shifts of a few degrees at 10hz and 100khz all the way to multiple shifts at different freqs all the way through the audio spectrum. Gear with transformers will give you more phase shifting thats less predictable than solid state designs. This is especially true of very colored mic pres. If youve ever tried running a track through a really colored transformer mic pre and then mixing it with the original, youll hear all sorts of strange phase artifacts. Thats part of the reason why these kind of designs sound so cool though.

Off the top of my head Id guess that a comp like the RNC, which has a very short solid state audio path (and isnt bandwidth limited, I think), would give you the least phase problems and something like a Mu compressor, especially older designs, will give you the most. This is because of high ratio step up and down transformers, and lots of caps as high and low pass filters in the audio path.

However none of this is relevant if you bring up the fader and it sounds good, which it clearly does to lots of you (and to me). Paralell processed tracks are rarely at equal level to the original anyway right? So youre limiting the audibility of phasing problems right there.


M

No, that's not what everyone's talking about. What they're referring to, I originally learned as "upward compression" if that helps at all...

and

That's something I was just considering today... It doesn't make much sense (to me) when you're going to 1's and 0's straight away, but if something's hitting the tape machine first and you want to keep the saturation and they dynamics of it (i.e. heavy guitars), then that seems like a great solution... Can't say I've tried it though, and I don't see myself doing any heavy guitars any time in the near future...

I still use the Valley People Dynamite & Gain Brain for this application.... unbeatable...