Question re MTurboReverb

[jmg8]

Don't worry I get it now, thanks.

[DarkStar]

I think that I misunderstood the % figures.

I was thinking that the output from the LR2 Out Mix was half the LR1 output and half the LR2 output (i.e. reduced to keep the volume level sensible). So it seemed sensible to add in half the LR2 output. I've been over-thinking it (not for the first time )

Now my understanding is that the LR2 Out Mix is made up of 100% of the LR1 Out Mix (itself being 100% LR1 output) PLUS 100% LR2 output. So 100% of the LR3 output should be added in too. The % just means how much of the signal is used in the mix Here's the edited pic.



I'm glad we did this - it's been an education, hasn't it

[MeldaProduction]

[spencerlee]

Vojtech Says "Ha, what??? If you want all stuff in parallel (as many as you want), just set input to 100% dry and output to 100% out, that's all, no black magic there "

Now this is why I gave up because that does not work.
Set all inputs to 100% dry 0% Prev and all outputs to 100% out 0% Prev.
Then set volumes on every Early and Late reflection to nothing.
Then send tone to the input and bring each volume up one at a time and you would expect to hear tone.
Wrong. You only hear tone out of Late Reflection 4.

Spencer

[spencerlee]

Then if you put the output to 100% Out and 100% Prev this gives you tone but there is 9db difference between ER1 and ER4 for the same signal input. So this does not work either.

I did suggest a solution on the first page that everybody would understand, then you would not need this thread.

[MeldaProduction]

Hehe, that's because I didn't check properly . It needs to be 0% prev 100% in and 100% out 100% prev to make it all parallel, and it's default. I'm not sure what you are talking about gain difference, please send me the settings. Sorry, but nobody needs a mixer.

[dmbaer]

Hehe, that's because I didn't check properly . It needs to be 0% prev 100% in and 100% out 100% prev to make it all parallel, and it's default. I'm not sure what you are talking about gain difference, please send me the settings. Sorry, but nobody needs a mixer.

If it's all parallel, the 100% in, 0%prev on input for all the late modules makes sense. On output, doesn't it just matter that all are set to 100% out. Should not the prev output settings be completely irrelevant since none of the late modules are taking prev as input? I'm not trying to split hairs here, just trying to truly understand how this thing works.

[MeldaProduction]

Not really. We actually discussed it recently with spencerlee, and the thing, that's probably causing the confusion is that it is technically all serial. If you look at darkstar's diagram, it's pretty clear there. Basically you can hardly simulate serial processing from parallel pipeline, but it is possible the other way around.

Now, you can see it this way:
Input controls how much the signal gets from the dry and how much from the current "reverb signal".
Output controls how much of the "reverb signal" stays there and how much output of this stage is added to it.

So if all inputs are taken just from dry and outputs are added to the reverb signal equally, you have parallel processing.
If all inputs are taken from current "reverb signal" and output from that stage is replacing the current "reverb signal", you have serial processing.

The confusion about gain is that it is all designed in such a way that it maintains energy, so that gain compensation isn't needed. The problem is that in the parallel case the compensation is not linearly spread between the stages, so that if you use say all 4 LRs, then LR1 and LR2 will be 6dB less loud than LR4, and LR3 will be 3dB less loud. You can certainly easily compensate for that using the volume, but that's what was confusing spencer.

[caulixtla]

The nice thing about this architecture is that it lets us have up to four tanks in multiple combinations of serial and parallel:

* Four in serial:

IN>LR1>LR2>LR3>LR4>OUT

* One in parallel, three in serial

IN>LR1>LR2>LR3>OUT
IN>LR4>OUT

* Three in parallel, first one two in serial

IN>LR1>LR2>OUT
IN>LR3>OUT
IN>LR4>OUT

* Four in serial:

IN>LR1>OUT
IN>LR2>OUT
IN>LR3>OUT
IN>LR4>OUT

The most useful configuration is probably something like this:

IN>LR3>OUT This would be early reflections, such as st[v;20[a(0.36);fl;fh];v;#[a(0.36)];v;#[a(0.36)]] which gives us a cushion of reverb while we wait for the main reverb to build up (length set to 100ms)

IN>LR1>LR2>OUT LR1 would be an input diffuser, such as a(0.7;11);a(0.7;23);a(0.7;9);a(0.7;19) (length set to 100ms); LR2 would be the main reverb tank, such as simply “r” or maybe swap;p[#an[swap;fl;a;fh;a;v;a]]

We use the early reflections — here called LR3 — to quickly give the reverb body. We use the LR1 as an input diffuser to smooth out the percussive transients in the input signal. LR2 provides the feedback loop for the main reverb tank.

Note that, if trying to sound like an actual room, the ER modules includes with MTurboReverb work a lot better for early reflections; using a LR modules for early reflections is best when trying to simulate the architecture of a 1980s digital reverb

Sean Costello, ValhallaDSP reverb guru, discusses this arrangement of reverb modules here: https://valhalladsp.com/2012/08/21/buil ... laubermod/

I know this is some complex stuff, but think of MTurboReverb as the Yamaha DX7 of the 2010s. We have had algorithmic reverbs since 1976, but this is the first time, in 40 years, we have had a programmable algorithmic reverb. Before MTurboReverb, algorithmic reverbs were preset only machines. If you wanted to change the preset, you had to hope the reverb designer included another algorithm (and the number of one-algorithm reverbs costing hundreds of dollars out there is amazing) — only Sean Costello of ValhallaDSP has added more algorithms to his VST reverbs for free — or you had to buy a different preset-only reverb.

With MTurboReverb, if you don’t like an algorithm, you can open up the hood and change it. Sure, most people won’t do this — just as most people didn’t program the DX7 in the 1980s. But, it allows third parties to make full reverb algorithms, allowing one to expand MTurboReverb’s sound far more than would be possible if only the developer could make a new algorithm. Just because MTurboReverb’s designer did not consider 1980s “gate reverb” programs important enough to implement doesn’t mean that MTurboReverb doesn’t have a “gate reverb” implementation; I spent one evening whipping one up so now, if you want to recreate the beginning of “Don’t Lose My Number” in MTurboReverb, you can. Chandler has made a great sounding spring reverb — it doesn’t matter what Melda Productions wants or has the time to implement — as long as the users want, say, a spring reverb, MTurboReverb will have one.

Edit Some technical errors corrected. Unedited version

[MeldaProduction]

Exactly!

[caulixtla]

In more detail:

IN>ER1>OUT
IN>LR1>OUT

LR1 Input: 100% In, 0% Prev Output: 100% Prev 100% Out
LR2, LR3, LR4: Turned off

This is nice in its simplicity. So far, all of the algorithms I have submitted use this simple configuration

Early/Late change the output balance from being 100% early reflections to being 100% late reflections

--

IN>ER1>LR1>OUT

LR1 Input: 0% In, 100% Prev Output: 0% Prev 100% Out
LR2, LR3, LR4: Turned off

Note that Early/Late has to be 100% (i.e. 100% late) for this to work

--

IN>LR1>LR2>LR3>LR4>OUT

LR1 Input: 100% In, 0% Prev Output: 0% Prev 100% Out
LR2 Input: 0% In, 100% Prev Output: 0% Prev 100% Out
LR3 Input: 0% In, 100% Prev Output: 0% Prev 100% Out
LR4 Input: 0% In, 100% Prev Output: 0% Prev 100% Out

--

IN>LR1>LR2>LR3>OUT
IN>LR4>OUT

LR1 Input: 100% In, 0% Prev Output: 0% Prev 100% Out
LR2 Input: 0% In, 100% Prev Output: 0% Prev 100% Out
LR3 Input: 0% In, 100% Prev Output: 0% Prev 100% Out
LR4 Input: 100% In, 0% Prev Output: 100% Prev 100% Out

--

IN>LR1>LR2>OUT
IN>LR3>LR4>OUT

As it turns out, this can’t be done (It’s a somewhat unusual configuration)

--

IN>LR1>LR2>OUT
IN>LR3>OUT

LR1 Input: 100% In, 0% Prev Output: 0% Prev 100% Out
LR2 Input: 0% In, 100% Prev Output: 0% Prev 100% Out
LR3 Input: 100% In, 0% Prev Output: 100% Prev 100% Out
LR4 N/A (This module should be turned off)

This is a common configuration, where LR1 is the input diffuser, LR2 the main reverb tank, and LR3 the early reflections.

LR3 Output changes the early reflection balance. 100% Prev 100% Out is a 50/50 mix of early and late reflections; 100% Prev 0% Out is pure late reflections, 100% Out is pure early reflections. By having the Early/late mix knob in Easy mode be connected to LR3 Output instead of early/late, it’s possible to have early/late balanced controlled when making an easy-to-use preset.

--

IN>ER1>OUT
IN>LR1>OUT
IN>LR2>OUT

LR1 Input: 100% In, 0% Prev Output: 100% Prev 100% Out
LR2 Input: 100% In, 0% Prev Output: 100% Prev 100% Out

[dmbaer]

Now, you can see it this way:
Input controls how much the signal gets from the dry and how much from the current "reverb signal".
Output controls how much of the "reverb signal" stays there and how much output of this stage is added to it.

Well why didn't you just say that in the first place?

Seriously, though, this finally is starting to make some sense and the above description really seems to be as simple a statement for which one could hope that explains how this works.

[spencerlee]

Hey Guys
I need some help.
I have this plate algorithm: p[2p[v;p[4c];p[10c];crosslp[f;f;p[10c]]];r]
I want to put a slight modulation into it.
How do I do this.
Any help would be great.
Spencer

[Chandlerhimself]

Hey Guys
I need some help.
I have this plate algorithm: p[2p[v;p[4c];p[10c];crosslp[f;f;p[10c]]];r]
I want to put a slight modulation into it.
How do I do this.
Any help would be great.
Spencer

You can use the LR modulation control, or use the ER modulation. You can also use V. Use the first parameter to adjust the amount. ex V(4) will be a noticeable vibrato. V(0.15) will be very subtle and should help break up some of the metallic sound without sounding like chorus.

[spencerlee]

I want to just modulate the delay line not the input signal.
V looks like what I want but where do I put it in the algorithm.
Spencer