Alesis midiverb/quadraverb emulation...

[sergiofrias]

Hello Sean,
i would love to see a emulation of the alesis large hall algorithms found on midiverb or quadraverb,i've seen your comments how your algorithms are similar to alesis,so is it possible to emulate the sound with valhalla room?i'm a fan of 90's trance music,a lot of producers in that time used Alesis reverb units. Cheers.

[Bronto Scorpio]

I'd like to know this too

Cheers
Dennis

[valhallasound]

The Early section of ValhallaRoom uses a signal processing technique that (as far as I know) isn't found in any "vintage" reverbs. Most older reverb, such as those from Lexicon, Alesis, Eventide and AMS, use several allpass delays in series feeding the input of the late reverb, in order to diffuse the input signal. This creates a distinctive metallic coloration with impulsive signals, that I wanted to avoid in ValhallaRoom. This coloration is also part and parcel of the sound of these older reverbs, so ValhallaRoom wouldn't be able to replicate this.

A few of the newer Dark Reverbs in ValhallaRoom (Nostromo/Narcissus/LV-426) take some pointers from Alesis designs. However, they aren't based on any specific Alesis algorithm, so much as discussions I had with Keith Barr before his untimely death.

In order to really dial in the Alesis sound, you would need to have Alesis-style algorithms, with allpasses feeding the inputs for diffusion. It would also be helpful to have a late diffusion control, for adjusting the diffusion of the later reverb. I think that the Quadraverb called these controls "Diffusion" and "Density," while older Lexicons had "Diffusion" and "Definition." The idea is that you can turn down the echo density of the late reverb, to get more defined echos in the early decay. ValhallaRoom doesn't have the controls for this, and I don't know if I could add them to all the algorithms.

I'm not saying never to emulating Alesis reverbs. I love a lot of music created with the Midiverb II and Quadraverb, especially artists on the Warp and Kranky labels. I'm just saying that ValhallaRoom would not be the proper vehicle for these older allpass-loop style reverbs, and that a different plugin "container" would be needed for these emulations.

ValhallaRoom was designed to avoid the artifacts of the older reverbs, and was very much the work of someone who had something to prove to the world. For my future work, I don't feel like I need to try to create anything "perfect" so much as "awesome." So future plugins may end up "embracing the grain" in order to emulate the spacious and clear qualities of the older algorithms.

I should also note that I probably wouldn't try to create sample-by-sample recreations of the existing Alesis algorithms. I think of the Alesis algorithms as a particular "school" of algorithm design, that is closest to the Lexicon 224, but with some simplifications that actually result in a clearer sound in the late decay. I would definitely haul out my Alesis boxes when dialing in the sounds (I have a Midiverb II, Quadraverb and Midiverb IV), but the goal would be to create Valhalla reverbs with an Alesis flavor, or Alesis-style reverbs with some Valhalla touches.

Sean Costello

[sergiofrias]

Thank you Sean for your response,i would love to see in the future some plugin that had the same alesis algorithms,maybe in the future you could develop one in your spare time .it would be great if Alesis make a plugin version.I don't have any midiverb or quadraverb but they are quite cheap,so maybe i should buy one or get impulse responses from them.

keep up the good work!!

Valhalla Room is amazing!!

cheers

Sérgio

[maxmace]

Let me +1 Sean's advice about hooking up a real Alesis box (when necessary). They are dirt cheap. I have a few QV, Wedge and Microverb II "spaces" that I still find useful. These older verbs also respond well to capturing their IRs, so that is also an option (if that doesn't offend your sensibilities).

My $.02 is that ValhallaRoom is insanely good at *not* sounding metallic, and avoiding all the pitfalls of the old algorithmic 'verbs that we *didn't* care for. Viva ValhallaRoom.

Sean, glad to hear that you knew Keith Barr. I think I remember you mentioning a couple of his tricks that allowed Alesis devices to perform so well (and musically) within the modest processing power of the era. He was a real pioneer.

[valhallasound]

Let me +1 Sean's advice about hooking up a real Alesis box (when necessary). They are dirt cheap. I have a few QV, Wedge and Microverb II "spaces" that I still find useful. These older verbs also respond well to capturing their IRs, so that is also an option (if that doesn't offend your sensibilities).

What algorithms/presets do you like on the Alesis reverbs? I don't have a Wedge, but I've heard that it is similar to the Midiverb IV.

My $.02 is that ValhallaRoom is insanely good at *not* sounding metallic, and avoiding all the pitfalls of the old algorithmic 'verbs that we *didn't* care for. Viva ValhallaRoom.

Thanks!

I am starting to wonder if there were some characteristics in those older reverb algorithms that were desirable, that were tied in with some of the "undesirable" traits. A few weeks ago, I did some tests on an AMS RMX16. Grainy as hell, totally metallic, impulse responses sounded HORRIBLE. Put it into the mix, and things were beautiful and 3D.

Sean, glad to hear that you knew Keith Barr. I think I remember you mentioning a couple of his tricks that allowed Alesis devices to perform so well (and musically) within the modest processing power of the era. He was a real pioneer.

I didn't really know Keith Barr. We had some conversations in the Spin Semiconductor forum, and had a few lengthy email exchanges. I was struck by what a nice guy he seemed to be - very down to earth.

Sean Costello

[maxmace]

What algorithms/presets do you like on the Alesis reverbs? I don't have a Wedge, but I've heard that it is similar to the Midiverb IV.

I have 2 particular ones that I go to:

(1) On the Wedge there is a preset called "Capitol Crooners" that is a great vocal diffusion patch. Not so much a plate (although that may be the algorithm), but more like an old echo chamber. Very warm.

(2) On the Microverb II there is an effects hall called "endless space" (according to the manual) that you have covered very well on the new LV-426 "Deep Space" preset. I use if for underwater environments and, uh, deep space. I am a supercollider expert, but for most transient sounds where you'd go granular/PaulStretch, a crazy FX reverb is the better, more musical choice (IMO).

I had other Microverb models, but there is something about the _Microverb II_ that made it head and shoulders above any other similar gizmo. For some reason it just sounds like a million bucks. The Quadraverb should have covered anything the Microverb II could do, but I can't coax the same sounds. I consider the *II* to be my secret weapon.

I never owned the Midiverb IV, but the Wedge was cheap when I needed it and it performs well. The box looks a lot like the SR-16 drum machine instead of a rack unit. I think the Wedge and the Vocal preset reverb they made (the one with the funny name) got bad reputations at the time, even though they were just the same chips as their flagship reverbs (in a different-shaped boxes).

[backtomusic]

...but the goal would be to create Valhalla reverbs with an Alesis flavor, or Alesis-style reverbs with some Valhalla touches.

Sean Costello

If this would happend -->

I always loved my Midiverbs and I had a hard time finding a software reverb(regardless of price) that I liked as much. ValhallaRoom was the first I liked and to this date there is only one other soft-reverb I like, but don't need since I got Valhalla.

But if I had a vote for a "unknown new valhalla produkt" or "valhalla verb with alesis flavor" i would vote for the unknown. After all I still got my Midiverb if needed it.

[sergiofrias]

Hello Sean,i remember from some of your talks with the late Keith Barr, you once mentioned that he replied in a email that Quadraverb used 4 parallel loops (2 x AP + 1 delay) ,outputs from delay taps...

Could midiverb/quadraverb used this algorithm?:



if so,would midiverb shared the same design?

thank you

Sérgio Frias

[kmonkey]

There's a preset called Bloom on my midiverb. I think it's number 49 and it is lovely I think i spent whole day patching my CR78 drum machine in to patch number 49..total trip LOL

edit: here are some examples - really like it http://www.synthmania.com/midiverb_ii.htm

such a class for such cheap and old device..amazing

Dry/Wet

XLargeWarm preset
http://www.synthmania.com/Alesis%20Midi ... %20sec.mp3

Deep chorus
http://www.synthmania.com/Alesis%20Midi ... us%201.mp3


Tap pan

http://www.synthmania.com/Alesis%20Midi ... %20Pan.mp3

[valhallasound]

There's a preset called Bloom on my midiverb. I think it's number 49 and it is lovely I think i spent whole day patching my CR78 drum machine in to patch number 49..total trip LOL

ValhallaShimmer can get kinda bloomy. Similar principle going on: lots of allpass delays in a row, with coefficients in the 0.5 to 0.618 range, so that the attack gets totally elongated. ValhallaShimmer won't emulate whatever fixed point horrors are going on inside the Midiverb II.

edit: here are some examples - really like it http://www.synthmania.com/midiverb_ii.htm

such a class for such cheap and old device..amazing

Dry/Wet

XLargeWarm preset
http://www.synthmania.com/Alesis%20Midi ... %20sec.mp3

That first example has SUCH A BAD SYNTH SOUND. Seriously. Horrible.

Turn on the Midiverb, and set the time machine to 1994 Warp records! Listen to how that crappy synth sound turns into beautiful rumbly goodness! Ahhhh...

Sean Costello

[valhallasound]

Hello Sean,i remember from some of your talks with the late Keith Barr, you once mentioned that he replied in a email that Quadraverb used 4 parallel loops (2 x AP + 1 delay) ,outputs from delay taps...

Could midiverb/quadraverb used this algorithm?:



if so,would midiverb shared the same design?

The algorithm in the picture is a later Keith Barr topology. It can be viewed as 4 parallel loops, as the input is inserted into 4 places (in front of each 2AP pair). However, the end of each loop is fed into the beginning of the next loop, so the feedback characteristics are much different. In practice, converting "parallel comb filters with embedded allpasses" into a single loop results in a much richer decay.

My guess is that the "single allpass loop" topology hadn't been figured out by Keith Barr for the original Midiverb. The Quadraverb used the 4 parallel allpass loops. By the time of the Midiverb IV and Wedge, the single allpass loop was definitely in use.

The impression I get from my communications with Keith Barr (and the code examples on the Spin Semiconductor site) is that the Alesis reverbs would use a LOT of topologies. Most of them consisted of allpass delays inside of some feedback configuration, but there was a great amount of variation with regards to the number of allpass delays, the number of inputs and outputs, the output taps, and the particular feedback paths. An allpass loop could have 2 or 4 input points, any number of output taps, 1, 2, or 3 allpasses in series, and any number of input diffusion allpasses. So the Midiverb II probably had a bunch of different reverb topologies for the presets, instead of a few basic topologies with different parameters.

Sean Costello

[valhallasound]

A great example of the Midiverb sound:

http://www.youtube.com/watch?v=6PIg4wcbk0w

The first minute or so is spent programming the sequence into the MC-202, so feel free to skip that. Once the reverb is turned on, it sounds RIGHT.

Sean Costello

[sergiofrias]

" However, the end of each loop is fed into the beginning of the next loop".

that's very interesting, when i was drawing the above topology i thought that the the output taps from the delays on each single loop were simply added for the output,i made the assumption that they would not feed into the next loop,if so,that explains a lot of that lush reverb tail.

My guess is that the "single allpass loop" topology hadn't been figured out by Keith Barr for the original Midiverb. The Quadraverb used the 4 parallel allpass loops. By the time of the Midiverb IV and Wedge, the single allpass loop was definitely in use.

i don't hava a quadraverb or midiverb to compare...how much the single loop topology improved the sound,did it increased density?wich one you prefer?,what are the differences (in sound)?

The impression I get from my communications with Keith Barr (and the code examples on the Spin Semiconductor site) is that the Alesis reverbs would use a LOT of topologies. Most of them consisted of allpass delays inside of some feedback configuration, but there was a great amount of variation with regards to the number of allpass delays, the number of inputs and outputs, the output taps, and the particular feedback paths. An allpass loop could have 2 or 4 input points, any number of output taps, 1, 2, or 3 allpasses in series, and any number of input diffusion allpasses. So the Midiverb II probably had a bunch of different reverb topologies for the presets, instead of a few basic topologies with different parameters.

I love to see your knowledge on those old topologies, yes ,it seems like the first midiverb was much more experimental,if only we had access to the algos...sadly Keith Barr isn't around to tell more about it,maybe it will remain a secret untill someone reverse engineer it,like warp69 did with relab lx480...

[valhallasound]

" However, the end of each loop is fed into the beginning of the next loop".

that's very interesting, when i was drawing the above topology i thought that the the output taps from the delays on each single loop were simply added for the output,i made the assumption that they would not feed into the next loop,if so,that explains a lot of that lush reverb tail.

The taps marked Lsum go to the left output, and Rsum go to the right output. The tap at the end of the delays (D3, D6, D9, D12) go into the inputs of the next 2AP/1-delay "block."

i don't hava a quadraverb or midiverb to compare...how much the single loop topology improved the sound,did it increased density?wich one you prefer?,what are the differences (in sound)?

The single loop has a much richer sound, for the most part, as each signal gets to travel through the entire loop before it is repeated. Much more thorough mixing. The parallel loops have smaller subsections repeating. I think that the MC202 demo reverb has parallel loops. I can hear some pretty obvious patterns, but it works for that material.

I love to see your knowledge on those old topologies, yes ,it seems like the first midiverb was much more experimental,if only we had access to the algos...sadly Keith Barr isn't around to tell more about it,maybe it will remain a secret untill someone reverse engineer it,like warp69 did with relab lx480...

There are enough Keith Barr written algorithms on the Spin Semiconductor site to keep people busy! Most of the other Alesis algorithms are far from "perfect," but they are certainly interesting. My guess is that, if you experiment with a bunch of allpasses in a loop, or a few loops, you can get close.

As far as strict reverse engineering, this probably won't happen anytime soon, primarily due to financial reasons. The Lexicon boxes get reverse engineered a lot, but the original hardware stills sell for $2K to $5K or more. The Alesis boxes can often be found for under $50. It takes a LOT of time to reverse engineer something by ear, even if you have a good idea of what the internals of the algorithm are, so there has to be a return on the investment.

Personally, I look at the Alesis stuff as inspiring ideas for future work, versus some "holy writ" that needs to be exactly copied. If you want the EXACT Midiverb II sound, it is still cheaper to buy the original box than, say, a hypothetical ValhallaMidiverbII.

Sean Costello