MTurboReverb preset making action, let's make the ultimate reverb! ;)

[Sampleconstruct]

Processing a simple Tremor beat with a tempo-synced MTurboReverb Preset where numerous parameters are being modulated by two synced step sequencers and an LFO modulates EQ frequency. Also randomizing algorithms on the fly later int he video - great fun, non-linear playground.

[caulixtla]

After discovering an unpleasant resonance in my “Gravity Well” algorithm, I generated an IR of the algorithm to find frequencies resonating too long. With this IR, I updated the algorithm to remove the resonances, as follows (the algorithm is one line in MTurboVerb, three lines here):

a(0.7;11);a(0.7;23);a(0.7;9);a(0.7;19);
f(505;peak;16;-12);f(1510;peak;8;-4.5);
swap;p[#an[swap;fl;a;fh;a;v;a]]

I made an IR of this algorithm to verify I removed the offending resonances. I have also put the IR on Soundcloud for people to download; this file can be used with convolution reverbs:

https://soundcloud.com/caulixtla/irgene ... 12-18-2-ir

It’s good to see MTurboReverb can make some far out stuff, but for me its main appeal is that I can make my own “conventional reverb” algorithms for it, allowing me to make an album with my own virtual “chamber” in it. I’m someone who thinks the more algorithms a reverb has, the better (as long as they sound good, which they can fairly easily via brute force with MTurboReverb, since I can put 162 allpasses and 18 delay lines in a reverb tank and run it at 96khz — for decreased latency — without it completely eating my CPU). It saves me having to wait for another reverb designer to either make a new reverb or to add algorithms to one of their reverbs.

For people who think MTurboReverb is too complicated: Don’t click on that edit button. I’m sure this reverb will have a number of third party presets; just treat it like a normal preset-only non-programmable reverb (i.e. any reverb made before MTurboReverb [1]) with a number of third party custom algorithms.

[1] With the possible exception of the Axoloti Core, but that device’s chip limits the kinds of reverbs it can have.

[Chandlerhimself]

It looks like everyone is coming up with cool algorhythms.

I'd just like to ask what everyone else is making. I've made a room preset and I'm almost done with a Spring preset. I'd like to try to make a random hall style preset and a more natural hall style preset if I have time.

What is everyone else making. It would be a shame if we had 100 different halls but only 1 plate. If someone else is already making a random hall I'll move on to something else. It seems like other people are already making fx style reverbs and ambient style stuff. I was going to try to make something like a blackhole, but I think other people have it covered.

If you're working on presets please write down what you're working on so others will have an idea of what other types of Algo we need.

[caulixtla]

I'd just like to ask what everyone else is making. I've made a room preset and I'm almost done with a Spring preset. I'd like to try to make a random hall style preset and a more natural hall style preset if I have time.

Naming algorithms is hard: https://valhalladsp.com/2014/01/18/nami ... lgorithms/

That in mind, I would say my submissions (which I just sent to Melda / Vojtech) consist of:

* A chamber reverb
* A large hall reverb (more or less the same algorithm as the chamber reverb, but this one has more sustain and a modulated tail)
* A non-linear reverb (sometimes called a “Gated” reverb)
* A room or small hall reverb (it’s a 10-character proof of concept but can sound good with some sources)

[chris.r]

Small rooms will be the main focus here. Maybe plate if I'll come with anything usable.

[MeldaProduction]

Chandler: Don't worry, statistically it's improbable . And anyways as cautixtla wrote, saying what exactly is hall and what is room is quite problematic .

Meanwhile folks, finishing new update, just for you beta folks, quite a few new modules here : XC (CircularComb compensated), Tremolo, Autopan, Pitch, FreqShift, Widening, LRtoMS, MStoLR, FDN. So that you have something to have fun with during Christmas

[Ah_Dziz]

Chandler: Don't worry, statistically it's improbable . And anyways as cautixtla wrote, saying what exactly is hall and what is room is quite problematic .

Meanwhile folks, finishing new update, just for you beta folks, quite a few new modules here : XC (CircularComb compensated), Tremolo, Autopan, Pitch, FreqShift, Widening, LRtoMS, MStoLR, FDN. So that you have something to have fun with during Christmas

Sweet Baby Jeebus! This Reverb is ridiculous!

[j.mayr]

getting better & better......
I´m glad, Vojtech added the pitch module for the "Shimmer" stuff!!!!

[caulixtla]

Meanwhile folks, finishing new update, just for you beta folks, quite a few new modules here : XC (CircularComb compensated), Tremolo, Autopan, Pitch, FreqShift, Widening, LRtoMS, MStoLR, FDN. So that you have something to have fun with during Christmas

OK, I have two feature requests which should be fairly easy, and a feature requests which would be fairly hard. Consider these optional; I can make nice sounding reverbs with the current version of MTurboReverb, but the following feature requests would give me more flexibility in designing algorithms.

The reasonably easy feature request:

To properly implement a classic Lexicon-style “figure 8” reverb tank (The part of a Late Reflection/LR that causes the reverb decay to happen), we need to have the ability to have the ability to 1) Input a signal in to the tank at any arbitrary point in the tank (This is not used as part of the feedback path in classic Lexicon topologies) 2) Output the signal directly out of the tank (not going back to the input as part of the feedback path) 3) The ability to lower the signal, depending on length (the longer the length, the less we lower the signal)

Here, “i” (input) and “o” (output) will take a single argument: A number (from -1 to 1; negative numbers invert phase) which is how much we multiply the signal by before adding it to the tank’s input or output. “ll” (level-length) will be similar to “l” (level), but it decreases the level more the shorter length is.

So, if we have something like an[v];o(1);d;a;fh;a;o(-1);d;o(1);ll;i(1);an[v];d;o(1);fh;a;o(-1);d that gives us a simple Lexicon-like “plate” reverb main tank (we would use ERs or another LR tank without feedback to diffuse the signal before it hits this tank).

(Observe the multiple places where we grab the signal from the middle of the tank to put in our output; the fact that the signal from before this tank is added to this tank in two different places; and how we, depending on the length, reduce the signal before adding this non-feedback input)

Also observe that the tank has only one feedback loop: The one MTurboReverb already has where we put the input in at the beginning of the tank, have output at the end of the tank, and take the signal at the end of the tank (and nowhere else) to put back in the beginning of the tank with the feedback amount determined by the “length” parameter.

(I might also point out that classic Lexicon topologies have only one reverb tank, not one tank per channel, and just input and output tap different channels at different points in the tank, but pulling that off would probably break a bunch of assumptions MTurboReverb’s code has)

This all comes from Dattorro’s classic “Effect Design Part One” paper.

The harder feature request:

How about the ability to add a fairly short noise convolution to the reverb tank? This uses more CPU than an allpass, but has the advantage of not ringing or adding metallic artifacts to the sound; it can diffuse the signal more quickly and with less coloration than multiple allpasses in series.

It could be like this: nfir(10); (“Noise finite impulse response, 10 ms long”). So, instead of having something like 8a;p[#an[fl;a;fh;a;v;a]] we could have nfir(10);p[#an[fl;a;fh;a;v;a]] which will not have the delay multiple allpass filters have; it also might reduce the flutter in the tail without us having to do a CPU-heavy modulation of the tank’s delay lines.

[Chandlerhimself]

Chandler: Don't worry, statistically it's improbable . And anyways as cautixtla wrote, saying what exactly is hall and what is room is quite problematic .

Meanwhile folks, finishing new update, just for you beta folks, quite a few new modules here : XC (CircularComb compensated), Tremolo, Autopan, Pitch, FreqShift, Widening, LRtoMS, MStoLR, FDN. So that you have something to have fun with during Christmas

Perhaps you're right. It looks like people are all kind of doing there own thing, so there shouldn't be too much overlap.

I like the sound of these new modules. I'm assuming pitch will us to do shimmer fx, which is something I wanted. I can think of some uses for autopan, widening, and LRtoMS. XC seems interesting and I'm looking forward to experimenting with it. FDN, I'm guessing is a different type of feedback delay network. I've been using fdn4 alot, so maybe this will be nice too. I'm looking forward to seeing all the different things that are possible. Just when I thought I was getting the hang of the modules that are already there.

[caulixtla]

Tremolo

Will the tremolo allow the use of a random/noise signal as its modulation source? A tremolo with noise as its modulation source is key to getting the Lexicon “Random Hall” sound.

[MeldaProduction]

Ok, folks, so you should all have received an email .

Cautixtla: I'm afraid "inserting" input / output is not really an option, longs tech story regarding optimizations. But actually that's what the SerialTap (st) is for . As for inserting input, you could probably do that using parallel module. If you have some schematics you'd like to realize, send me an image, I may be able to quickly figure out the optimal approach. The thing is, in many cases there are multiple ways to do the same thing, or similar one.
Personally I'm not a fan of neither of these things, because from the technical point of view inserting input/extracting output doesn't really add anything new to the game. It only increases the chance of the metallic resonances. I understand it from the historical point of view, but it sort of seems obsolete.

Short convolution: Well, the thing is, when it is short it actual is ringing, slightly. We could add that but I don't see too much of a point. It is easy to build a well working decay to noise via CC, R, FDN and others, so doing to convolution seems like quite a waste of resources really. Btw, the 10ms example of yours - 10ms itself is actually 100Hz, so convolving with such a noise fragment could still ring in similar frequencies.

FDN: It is actually a generalized feedback delay network, pretty cool, and it's not as slow as I was afraid of, but not superfast either (though the room for further optimizations is close to none...).

White noise tremolo: Hmmm, it uses just classic sine. We could probably make a few more shapes, including noise, didn't think about that. I wonder how that could be used in a reverb, I mean, tremolo is just a ring modulation basically and with noise, well, hard to say how that would sound

[Sampleconstruct]

Wow, what an amazing update, with the new frequency and pitch shifting, this thing now can produce alien voices and things I've not heard coming out of any effect plugin I've used in my life.

So here I am experimenting live on video with a custom preset in MTurboReverb using two ER modules in serial, the first one using frequency and modulated pitch shifting. Input source is one of my Serum sound.

[caulixtla]

I have, after going through a bunch of metallic sounding reverb tanks, found something with very little metallic ringing:

p[#[a;d;a]];swap;fh;an[an[a;v;a;fh];an[2a]];a(0.7;11);an[a;fh];a

I think I can build a nice reverb with this (there are some resonant nodes which I might have to tame with a f() or two).

[MeldaProduction]

Wow, what an amazing update, with the new frequency and pitch shifting, this thing now can produce alien voices and things I've not heard coming out of any effect plugin I've used in my life.

So here I am experimenting live on video with a custom preset in MTurboReverb using two ER modules in serial, the first one using frequency and modulated pitch shifting. Input source is one of my Serum sound.