The case for delay after reverb for studio recording. Please add your experiences.

[Orvillain]

Lot of cats and ways to skin them.

 

[Mongillo19]

Look what you've done @JiveTurkey

 

[DLC86]

An FDN isn't inherently an LTI. It would not be an LTI if the feedback matrix or gains are modulated over time; which is common in reverb design, and you probably would struggle to find an FDN based reverb that didn't do that.

As I said I don't know how those work exactly, but what time are you referring to when saying "modulated over time"? Absolute time (so like an lfo attached to the level of each delay line) or the time over which the gain/feedback of each delay line is attenuated for each sample to create a given decay time? Cuz the latter would be still LTI and no different than what happens in the time domain of an IR... unless that gets changed in each sample somehow.

You should probably checkout some reverb algorithm designs, because more or less, your assertion is kind of laboratory test conditions. But in the real world, a lot of your assumptions fall apart instantly.

Do you know if Fractal reverbs use FDN? Cuz, as I said, they null perfectly when turning off distortion/modulation/compander/pitch shifting.
And what you mean exactly by "real world"?

I hear significant differences in your clips. As to whether I could specifically tell you what the order was in each of them, I can't. There is not enough context. Much in the same way I can tell you that a real amp versus Kemper profile has differences in the sonics, but I might not be able to tell you which is which.

The fact that differences are observable is enough to show that the differences are important.

There are "significant" differences but you're not able to tell which is which? This is not a comparison between some arbitrary amp profiles you know nothing about, @MirrorProfiles described quite accurately what happens when the modulation is applied after the reverb and it is quite evident in his sample.
So if these differences are "significant" to you, you should be able to tell which is which pretty easily just like in his sample, otherwise the difference is negligible and you've been arguing over my statement just for the sake of grinding my glans.

 

[riffy]

I will usually use either a multi tap delay or at least a ping-pong delay. To my ears, it does make a difference if it is before or after reverb. I prefer it before reverb when I decide to use reverb.

I tend to go a little overboard with reverb, and I’ve really had to cut back on it lately because it just muddies up the solo sounds, at least in my ears. It seems like my low mids and mid range just get Tubby and have no definition.

 

[DLC86]

IMO the modulation was more than wide enough in your examples but the reverb tail needs to be sufficiently long and/or loud to hear the modulation affecting it.

The reverb was set with mix at 50% and input gain at 50%, so roughly 6dB lower than dry signal, louder than what I'd tipically use if I'm not looking for an ambient tone.
Length was set to 3 seconds, so not very long but not short either... I'd say a good middle ground.
It seems you need specific settings to be able to discern it?

Here we go with the sweeping generalisations again. I think again, it just depends. I’d imagine more often than not the results they get and use are as a result of putting the pedals in a specific order. I’d imagine most people do pay attention to it, even if it doesn’t always make a noticeable difference.

I'd imagine most people put the reverb after instead cuz that's always been considered the "right" order and they imagine there's a difference even when there's none. Not everyone is obsessed with little details in their tone (and I'm saying it as someone who IS obsessed).

Similarly, IR’s might be good enough or even indistinguishable for a lot of reverb uses (at certain settings), but that doesn’t mean they don’t usually sound like ass. and similarly, I want delay to have non linear behaviours, and not to be reduced down to something without quirks. Quite often the reason we like particular bits of gear is because of what they do when they’re pushed near their limits and they show off some kind of unique/distinctive character.

Either way - I’m not seeing a strong case for the benefit of being indifferent to the order of reverb and delay. It matters, it makes a difference (even if you can’t always hear it).

You're telling that to someone who loves modulation and dirt in delays and usually even prefers to run them before the amp for that reason. But that has little to do with the perceivability of the difference between pre or post reverb, and aknowledging when the difference is minuscule doesn't mean one should be indifferent to the order either, quite the opposite actually. Here's a little background story:

I decided to make that video cuz, during a live stream with Marco, I said that in some cases it's better to run the delay after the reverb, for example when using a ping pong or panning delay, cuz reverbs are "mono in" on the Fractal and you'll always hear the reverb on the repeats in the center if you put the delay before it.
Another example is the 2290 which has L and R channels out of phase and the repeats won't produce any reverb if it is placed before, it's just like running them in parallel.

A lot of people in the chat simply could not accept this statement and wrote that putting the delay after the reverb is unnatural and sounds bad, so I wanted to prove them that in most cases the difference in tone given by the order is negligible and putting them the other way around can have far more important advantages in those cases.

After I made the video a bunch of those people wrote me to tell me they tried for the first time to invert the order and that I was right, with their typical setting they couldn't hear the difference but finally could make the reverb follow a ping pong delay or work properly with a 2290.
Other people instead (probably the majority) still insisted on saying they heard a difference even in cases where both blocks were LTI and nulled perfectly.
You just can't win against the power of suggestion

 

[MirrorProfiles]

It seems you need specific settings to be able to discern it?

Yes, I think that has been said along by both of us.

Regarding the rest, there is something pretty important being overlooked about the order, and essentially the reason why I said anything at all.

Even at settings where the difference is hard to perceive, you’re never far from adjusting settings to something that will give unwanted results.

The only time I’d run a reverb before a delay is when I purposely want to hear the reverb being affected by whatever the delay is doing. And vice versa (otherwise I’d run things in parallel).

It doesn’t really matter if it’s always perceivable - the goal is to avoid the unwanted results even having the opportunity of coming up, which means you can use either effect freely. You don’t have to worry about a reverb summing inputs to mono, or a delays stereo effect being compromised, or the reverb tail being modulated/being companded/reduced in bandwidth.

 

[laxu]

Has been like that for me for a long time, but by now I like the kinda mellowing effect a reverb has on some delay repeats.

That's what I like about the spring reverbs on the Strymon Volante and El Capistan. They can either smooth the tone or give it a bigger sound.

 

[Orvillain]

As I said I don't know how those work exactly, but what time are you referring to when saying "modulated over time"? Absolute time (so like an lfo attached to the level of each delay line) or the time over which the gain/feedback of each delay line is attenuated for each sample to create a given decay time? Cuz the latter would be still LTI and no different than what happens in the time domain of an IR... unless that gets changed in each sample somehow.

Here's a video for you:

Do you know if Fractal reverbs use FDN? Cuz, as I said, they null perfectly when turning off distortion/modulation/compander/pitch shifting.
And what you mean exactly by "real world"?

I don't know tbh. You'd have to ask Cliff. He did say in a prior thread that it was a similar approach to what Fractal use (in reference to a video by Signalsmith Audio on reverb design - and a central part of that was FDN's separated from a diffusion network)

There are "significant" differences but you're not able to tell which is which? This is not a comparison between some arbitrary amp profiles you know nothing about, @MirrorProfiles described quite accurately what happens when the modulation is applied after the reverb and it is quite evident in his sample.
So if these differences are "significant" to you, you should be able to tell which is which pretty easily just like in his sample, otherwise the difference is negligible and you've been arguing over my statement just for the sake of grinding my glans.

Well the comparison to amp profiles versus real amp is 100% a valid comparison. Ed has better ears than me and better language to describe what he's hearing. I'm just a lowly guitarist and paper shufflerer. Nowhere near on your guys' level of understanding.

 

[DLC86]

Here's a video for you:

Actually you mentioning this stuff caused me a sleepless night reading a couple papers on the subject... But I'll watch the video too.
I can't fully grasp the math behind it (last time I used matrices was almost 20 years ago) but from what I've gathered till now, gain/feedback/matrix modulation is commonly used to create, well, modulation, to make the reverb smoother and less metallic or, in some very specific cases, to simulate the movement of the source or the listener, but the latter is used only on very few plugins aimed to binaural rendering or atmos/ambisonics processing afaik, not something you'd find on a PCM 70.

The fact is that the reverberation of an ambience captured by a still mic is pretty much LTI (if we ignore distortion introduced by the mic/preamp, tiny variations in the air temperature and unless you have a room with floating walls and ceiling), so any algorithm trying to simulate recordings of real spaces can't depart too much from that.
Modulation is added to reverbs also cuz it resembles what we hear when we move across a room and we like that cuz we're never perfectly still in space.

Anyway, just to be clear once again, no one ever stated that reverbs can't be non-linear... the point was if those subtle non-linearities make a hearable difference when inverting the order with delay.

 

[Orvillain]

If the point is that sometimes the order of delay and reverb can be subtle enough to not really matter that much, then I think we are all agreed.

My interest in it was to try to answer the question of whether the order could never matter, and the obvious answer is no.

If the other related question is, does the order always matter? Then the obvious answer is also no.

Sometimes an FDN is LTI, and sometimes not. Sorry you lost some sleep! Consider it revenge for that time you travelled back in time and got me drunk enough to agree to making babies with my wife, coz now every night is like that for me

 

[ian_dissonance]

If your settings are low enough for the order to not matter does it even matter if they’re on?

 

[Sascha Franck]

Modulation is added to reverbs also cuz it resembles what we hear when we move across a room and we like that cuz we're never perfectly still in space.

Well, modulation (and other things) are also added to simply make it a more interesting effect. Same for delays.
I mean, pretty much no sane person is after realistic sounding spatial FX when running a guitar rig. What you usually want is an impression of width and depth, possibly spiced up with some other niceties. Because of that, delays/verbs can be modulated, saturated, "shimmered" and what not.

 

[Sascha Franck]

If your settings are low enough for the order to not matter does it even matter if they’re on?

Well, you can have 100% wet delay/reverb combinations without the order making any difference. It's really just about the type of delay/reverb.

 

[FuzzyAce]

So order doesn't matter unless it does. Glad it's been cleared up, it only took 5 pages!

 

[Orvillain]

Well, modulation (and other things) are also added to simply make it a more interesting effect. Same for delays.
I mean, pretty much no sane person is after realistic sounding spatial FX when running a guitar rig. What you usually want is an impression of width and depth, possibly spiced up with some other niceties. Because of that, delays/verbs can be modulated, saturated, "shimmered" and what not.

Yeah that was kind of my point with the modulated FDN's too. I believe that often enough there is internal modulation that the user doesn't have control over, which simulates the characteristics of a real room.

The modulation 'added' to a reverb at a later stage is not the same kind of thing. Again, happy to be corrected. But I think there's two general modulation types to consider - realism modulation, and then additional creative modulation.

I might upload a few clips of an FDN reverb I did in Python as an experiment. See what you guys think of it!

 

[DLC86]

My interest in it was to try to answer the question of whether the order could never matter, and the obvious answer is no.

And no one ever said that in this thread in fact.

If the other related question is, does the order always matter? Then the obvious answer is also no.

No one ever said that either... But based on the background story I wrote earlier, there are a lot of people out there who are quite sure of that, that's why I brought this up to begin with.

If you lazy asses just spent 10 minutes watching my video (and making me rich with youtube's 0.002$ ad revenue) where all this was pretty clear, you would have saved us from a two days long discussion!
But it was (kinda) fun and informative anyway...

Sometimes an FDN is LTI, and sometimes not. Sorry you lost some sleep! Consider it revenge for that time you travelled back in time and got me drunk enough to agree to making babies with my wife, coz now every night is like that for me

Hey, don't blame me for your alcoholic spermatic incontinence!

 

[laxu]

Yeah that was kind of my point with the modulated FDN's too. I believe that often enough there is internal modulation that the user doesn't have control over, which simulates the characteristics of a real room.

The modulation 'added' to a reverb at a later stage is not the same kind of thing. Again, happy to be corrected. But I think there's two general modulation types to consider - realism modulation, and then additional creative modulation.

I might upload a few clips of an FDN reverb I did in Python as an experiment. See what you guys think of it!

Modulation and pitch shifting can also be added at different points. They could be added at the input, or only on the reverb trails (regeneration).

E.g Strymon Nightsky lets you choose which for its pitch shifting and they can sound quite different.

Timestamped to the voice section, at the end of it you can hear how it sounds and then the next section shows the regen version.



That pedal also blurs the line between delay and reverb as the "sparse" algorithm can sound like a delay if you just quickly strike a note, but becomes more reverby if you hold the note.

 

[Orvillain]

And no one ever said that in this thread in fact.

I did ask you to clarify what you were saying, and so far as I can tell you never did???

Hey, don't blame me for your alcoholic spermatic incontinence!

 

[JiveTurkey]

If your settings are low enough for the order to not matter does it even matter if they’re on?

 

[Sascha Franck]

Fwiw, U-He's MFM2 offers options to create rather complexed delay feedback networks, you can get some quite convincing reverb-ish sounds out of it. And then some, as there's plenty of things you can insert in the feeback path. Amazing plugin, really.