#2 Diving Deeper into Parallel processing and Phase

[AcousticDreams]

Recently I read a couple of articles that stated that some plug in's when used in parallel could create Phase issues.

Quote:

"if they incorporate oversampling because many oversampling methods are not linear phase.
Most eq and most compressor plugins shouldn't have an issue. Anything with distortion, anything that is known to be oversampled and anything with a non-linear phase crossover should be suspect of having a mangled phase response. Stuff I know to have mangled phase response - Waves C4, Waves C6, Wave RDe-esser (I think the regular de-esser is fine), Brainworks limiter, Decapitator, Guitar Rig (HQ mode is oversampled), most other guitar amp sims, Fabfilter Saturn. I've noticed that decapitator can mangle the HF response when in parallel, but it's not really noticeable if it's pushed hard. That said, it's fun on vocals."

Let us ask the question first: What is the real difference between Parallel processing a plug in, versus directly adding the plug in to your channel?

I was under the assumption that the only difference was that you were enabling the fader of the buss to become a wet/dry control in essence. Using the master as the dry and the buss as a wet signal. As not all plug in's, analog gear has a wet/dry mix as a built in feature.

I love having a Wet/Dry control. One of the reasons why I choose my Buzz Velox compressor for tracking. However, the second question I have is to make it clear if there are Other advantages to Bussing for Parallel wet/dry control? If say, the plug in already has a wet dry control such as most reverbs do?
And lastly, what do they really mean by oversampling in a plug in? Aren't all plug in's sampling? What is the difference between sampling and oversampling?

This recent bit of information comes at a timely manner, as I have but three more days in which to Purchase a "Decaptitator" from Sound Toys, at the substantially reduced sale price.

[jim1960]

Originally Posted by Knives&Guitars

This recent bit of information comes at a timely manner, as I have but three more days in which to Purchase a "Decaptitator" from Sound Toys, at the substantially reduced sale price.

Kazrog True Iron is also a very cool saturation plugin and it's also on sale. You might want to demo both.

[KevWind]

Knowing only enough to be dangerous this is just my method and thinking .

I don't know if or how "oversampling" might effect phase.
I assume many plugins use oversampling to avoid aliasing errors ?

I use both parallel and direct processing in my sessions (depending on the actual function of the processing) more on this in a bit.

For me the "real difference" between parallel processing, and direct on the audio track itself :::

#1 the ability to run multiple tracks into one instance of processing

#2 because I have both a Hardware Reverb and a Hardware Compressor, and I have them set up as inserts it (which I can use just like plugins) having them in parallel allows me to run multiple tracks into them simultaneously and is a key function in my mixing process.

#3 in a parallel processing configuration,, the signal coming from the outputs of the audio tracks are 100% dry going to the main outs and retain all the transient attack of the dry recorded signal ,, which is then added to (and is in addition to) the processed signal coming from the output of the parallel track (and from which the transients will be more diffuse because of the processing ). Which is why many engineers prefer parallel especially for time domain processing like Reverb,,and do not put reverb directly on the audio track's especially with multi track productions.

#4 as you say in a parallel configuration the fader/level of the send bus,,,, and or parallel channel fader,,,, then becomes the control for the wet to dry ratio (and thus amount of perceived effect) hitting the main outs

#5 Now for me another big advantage of parallel for Reverb specifically , is that I can EQ the reverb specifically for Reverb, without that eq setting effecting the audio track/s

In my productions I always have EQ (if only a high pass filter) on the audio tracks themselves or their sub bus and often a compressor as well . But I always do reverb in parallel. (And my 2 Bus compressor is in parallel also,, which all tracks come through going to the mains ....Sometimes I will put a tape delay plug in directly on an audio track specifically for a perceived effect.

[jim1960]

My preference for applying reverb in parallel is pretty simple: I think it's easier to dial in what I want when I have separate control over the fundamental (the track without reverb), for lack of a better term, and a fully wet track.

I'm not sure how much difference there is in doing it one way vs the other. I was taught early on to apply reverb in parallel and I've never seen a mixing engineer of any import ever say it's better to apply reverb directly onto a track, so there's been no reason for me to question the parallel standard.

In the history of mixing, there may have been a time when applying reverb in parallel wasn't an option so if someone was doing it that way back then... well, you work within the limitations of the gear you have. I'm neither seeing nor hearing about any advantage in applying reverb directly to a track. And since setting up a parallel reverb is a simple and quick process, there's no good reason for me to do it any other way.

[KevWind]

Originally Posted by jim1960

My preference for applying reverb in parallel is pretty simple: I think it's easier to dial in what I want when I have separate control over the fundamental (the track without reverb), for lack of a better term, and a fully wet track.

I'm not sure how much difference there is in doing it one way vs the other. I was taught early on to apply reverb in parallel and I've never seen a mixing engineer of any import ever say it's better to apply reverb directly onto a track, so there's been no reason for me to question the parallel standard.

In the history of mixing, there may have been a time when applying reverb in parallel wasn't an option so if someone was doing it that way back then... well, you work within the limitations of the gear you have. I'm neither seeing nor hearing about any advantage in applying reverb directly to a track. And since setting up a parallel reverb is a simple and quick process, there's no good reason for me to do it any other way.

For a solo acoustic guitar production I doubt it makes much difference
But for multi track productions ::::
The difference became readily apparent when I was talking on line classes at Berklee. We had a 35 to 40 raw recorded tracks of a rock production we were supposed to mix. And for time domain FX (both reverb and delay) we were instructed to experiment with both directly on the tracks and in parallel. What was a stark contrast was that all (everybody in the class) the sessions with directly on the tracks all sounded like mush , compared to the sessions where we used parallel

[Bob Womack]

We are blending some concepts and terms here and it can cause confusion.

The two basic methods of effect feeding are insert or send/return.
The Insert method inserts effects into a channel or bus via an insert point that contains an insert out that breaks out the channel's signal and allows it to be sent to the effect and then allows the effect's output to be inserted into the channel at the same point in the channel. Blending of the effect and clean signals happens within the effect module (aka plugin).
The Send/Return method designates a effects bus and a return channel. The channel signal is sent to the effects bus and that bus feeds the effects device (aka plugin). The output of the effects device's unmixed effect is returned to the mix via a return channel and blended into the mix via a fader.

Where the heck did all this come from? Originally, all effects had no ability to blend clean and effected signal within them. Chambers, tape delays, plates, gold foils, and the original digital delays (such as the Lexicon Delta T) had no blending capability and were extremely expensive. As a result, a send/return scheme on the console was necessary to blend them with a clean signal. As a result of their expense, studios had limited examples of each effect available so multiple sources often shared effects. Before about 1978 and the SSL SL4000 console, if you wanted parametric EQ or compression on a channel, you patched an expensive outboard parametric EQ or compressor into the insert point of a channel or bus.

Notes: If you decide to send/return to an effect (plugin) you need to set the effects module's master blend control to 100% effected or sending will alter the channel's level in the mix. Surprises will ensue.
Parallel effects such as parallel compression can only be accomplished via a send/return scheme BUT that scheme can be accomplished via the console OR, if the effects device (plugin) has facilities for it, within the effects device itself as an insert.

An historic/pragmatic note: If you are beginning in mixing, you may find it easier to route similar sounds, such as all voices and electric guitars, to a single send/return reverb, rather than trying to keep up with a zillion individual reverbs. Once you get comfortable you can start adding individual reverbs via insert.

Bob

[jim1960]

Originally Posted by KevWind

For a solo acoustic guitar production I doubt it makes much difference

Maybe not but I don't know that for sure. And since I never do solo acoustic, I'm not likely going to know in the future.

Originally Posted by KevWind

But for multi track productions ::::
The difference became readily apparent when I was talking on line classes at Berklee. We had a 35 to 40 raw recorded tracks of a rock production we were supposed to mix. And for time domain FX (both reverb and delay) we were instructed to experiment with both directly on the tracks and in parallel. What was a stark contrast was that all (everybody in the class) the sessions with directly on the tracks all sounded like mush , compared to the sessions where we used parallel

That goes right to a question I've been wondering about as I'm reading through this thread and the other similar thread... does applying reverb to the track directly weaken the fundamental of the track in a way that does not happen when applying reverb in parallel? I suspect it does but may not be as apparent on a solo acoustic track because you don't have the build up of a weakened fundamental going across numerous tracks. I could just as easily be wrong on that front though.

[KevWind]

Originally Posted by Bob Womack

We are blending some concepts and terms here and it can cause confusion.

The two basic methods of effect feeding are insert or send/return.
The Insert method inserts effects into a channel or bus via an insert point that contains an insert out that breaks out the channel's signal and allows it to be sent to the effect and then allows the effect's output to be inserted into the channel at the same point in the channel. Blending of the effect and clean signals happens within the effect module (aka plugin).
The Send/Return method designates a effects bus and a return channel. The channel signal is sent to the effects bus and that bus feeds the effects device (aka plugin). The output of the effects device's unmixed effect is returned to the mix via a return channel and blended into the mix via a fader.

Where the heck did all this come from? Originally, all effects had no ability to blend clean and effected signal within them. Chambers, tape delays, plates, gold foils, and the original digital delays (such as the Lexicon Delta T) had no blending capability and were extremely expensive. As a result, a send/return scheme on the console was necessary to blend them with a clean signal. As a result of their expense, studios had limited examples of each effect available so multiple sources often shared effects. Before about 1978 and the SSL SL4000 console, if you wanted parametric EQ or compression on a channel, you patched an expensive outboard parametric EQ or compressor into the insert point of a channel or bus.

Notes: If you decide to send/return to an effect (plugin) you need to set the effects module's master blend control to 100% effected or sending will alter the channel's level in the mix. Surprises will ensue.
Parallel effects such as parallel compression can only be accomplished via a send/return scheme BUT that scheme can be accomplished via the console OR, if the effects device (plugin) has facilities for it, within the effects device itself as an insert.

An historic/pragmatic note: If you are beginning in mixing, you may find it easier to route similar sounds, such as all voices and electric guitars, to a single send/return reverb, rather than trying to keep up with a zillion individual reverbs. Once you get comfortable you can start adding individual reverbs via insert.

Bob

Great explanation as usual , thanks
Yes to clarify I was, as you note, using the terms "direct" and "parallel" to simply distinguish between having the effect on the audio channel itself (direct)----- and having the effect on a return/aux channel (parallel) which I was assuming would be understood to be accomplished via a bus/send from the audio channel/s.
And good clarification point because I had just assumed that it was understood that with direct, one would be dialing the mix % control to a blend of wet/dry,,,,, and that in parallel the control on the effect would be 100% wet

[KevWind]

Originally Posted by jim1960

Maybe not but I don't know that for sure. And since I never do solo acoustic, I'm not likely going to know in the future.



That goes right to a question I've been wondering about as I'm reading through this thread and the other similar thread... does applying reverb to the track directly weaken the fundamental of the track in a way that does not happen when applying reverb in parallel? I suspect it does but may not be as apparent on a solo acoustic track because you don't have the build up of a weakened fundamental going across numerous tracks. I could just as easily be wrong on that front though.

I think perhaps (not sure about weaken the fundamentals) but I think it does does make the transients less detailed and reduces the attack and makes it more diffuse, which on multi tracks adds up.

Because to my thinking placing a reverb directly on the audio track is a bit subtractive (for lack of a better term) it takes away some transient attack and by adding diffusion (now arguably that can be compensated for some with pre delay) But in parallel ,, I view it as additive ,, 100% dry transient attack and then the more diffused reverb effect is added to that.sound in the overall mix . ...... But again I am guessing it makes little difference on solo acoustic. Or OTOH I could of course be full of horse feathers

[Doug Young]

You know, the thing is that this stuff is all software now. I'd say sonically, there *should* be no difference between inserts (with a blend capability in the plugin) and send/return as a general thing. It's all just combining 1s and 0s in the end.... But, every DAW and every plugin is written by different programmers, and there's no way to really know, short of measuring and listening, what bugs, design features or misfeatures any given piece of code has at this level.

To make it more complex, some plugins try to emulate the behavior or mis-behavior of hardware, and some at least attempt to include non-linearities that may have existed. So if a piece of old hardware was known to introduce phase changes, or unintended distortion, or noise, or..., the software developer may have tried (and succeeded or failed) to mimic that in the code.

Also most DAWs support plugin latency compensation, so they try to make sure that at least processing time is accounted for in different paths. In Logic, at least you can hear this. Activate as new plugin on one track of a mult-track recording, and you hear delay/phase issues for a second or two, until Logic does the calculations and corrects for it.

[jim1960]

Originally Posted by KevWind

I think perhaps (not sure about weaken the fundamentals) but I think it does does make the transients less detailed and reduces the attack and makes it more diffuse, which on multi tracks adds up.

Because to my thinking placing a reverb directly on the audio track is a bit subtractive (for lack of a better term) it takes away some transient attack and by adding diffusion (now arguably that can be compensated for some with pre delay) But in parallel ,, I view it as additive ,, 100% dry transient attack and then the more diffused reverb effect is added to that.sound in the overall mix . ...... But again I am guessing it makes little difference on solo acoustic. Or OTOH I could of course be full of horse feathers

I think we're talking about the same thing but you've found the more eloquent way to phrase it.

[runamuck]

Regarding using an effect as an insert or a send, I've found with software I can get exactly the same sound from one to the other but different settings are required.

[jim1960]

Originally Posted by runamuck

Regarding using an effect as an insert or a send, I've found with software I can get exactly the same sound from one to the other but different settings are required.

Exact same to your ears or did you null the tracks?

[Bob Womack]

Originally Posted by Doug Young

Also most DAWs support plugin latency compensation, so they try to make sure that at least processing time is accounted for in different paths. In Logic, at least you can hear this. Activate as new plugin on one track of a mult-track recording, and you hear delay/phase issues for a second or two, until Logic does the calculations and corrects for it.

Believe it or not, this was not always the case. My company got in on the ground floor of DAWs. At that time, only one DAW of all of them, Neundo, had built latency compensation into their architecture. ProTools went YEARS without that, and the panning of stereo mixes "in the box" would "fold up" and become more and more mono as you added processing due to phase issues. Pros ran multiple outputs to a console and mixed with the automation on the console, as if ProTools was just a big, complicated multitrack. Eventually, ProTools chose a point to re-tool from the ground up, include latency compensation, and declare that they would no longer be backwards compatible.

And that is why my company chose to go with Nuendo way back in 2001, as I remember. That and superior video handling at the time when we invested. Of course, ProTools grew up and became the largest pro platform on the planet.

Bob

[Doug Young]

Originally Posted by Bob Womack

Believe it or not, this was not always the case. My company got in on the ground floor of DAWs.

Yeah, I should have said "most modern" DAWs. It's pretty amazing to think back on how many features we take for granted today didn't exist not that long ago. That may be where some of these warnings Victor's reading about come from, "historical" (as in a few years ago) issues. I'm not sure what the state of DAWs like Audacity and Reaper are when it comes to stuff like this, either. Maybe still an issue?