Reverb Subculture

[seancostello]

Originally Posted by zmix

Hi Sean,

Have you ever listened to a Bode Frequency shifter?

I have been programming these since 1999 - it was one of my first goals in learning digital audio. I've done a lot with them since then, including putting them in the feedback path of allpass loop reverbs. I didn't like the results that much, at least as far as producing a modulated reverb sound.

There is a fair example in the Logic built in plugins calld "Ringshifter" if you set the modulation oscillator to 1.5hz or so the effect is very close to the what the modulation on the PCM96 "Hall" and "Random Hall" sounds like.

Here's a way to test things:

Run sine waves of different frequencies (100 Hz, 500 Hz, 1000 Hz, 2000 Hz) through the algorithms in question. See if the beat rates sound similar. If the beat rates are similar, then the modulation is some form of amplitude modulation (which frequency shifting can be classified as in some weird sense), which produces a beat rate that does not vary with input frequency. If the beat rates get higher with the input frequency, it is probably delay modulation.

The chorusing LFO in the PCM96 sounds like a square wave and has a maximum rate of 1.5hz. It is not randomized.

This sounds like the modulation produced by a triangle wave LFO, which produces an up/down modulation when used to modulate a delay line. In most digital systems, a square wave LFO ends up producing different delay lengths with no pitch change, except for a glitch when changing between the LFO output values.

Of course, this is based on one method of calculating delay modulation, while other methods could result in a square wave LFO producing the up/down modulation, and a triangle wave sounding more like a sine wave. It all depends if you are adding the modulation as an offset to your delay address, or doing something else. For BBD-based effects, the delay modulation LFO is usually controlling the clock rate, so a square wave LFO can end up producing an up/down pitch change.

Sean

[zmix]

Excellent points, Sean.

The modulation artifacts I'm demonstrating here may not be phase shift networks but the tonality is similarly disturbing. I just cannot believe that these algos made it to final release... When the original EMT 250 and 224 were developed the sheer man hours involved in listening and tweaking must have been boggling.!

[Casey]

Originally Posted by zmix

When the original EMT 250 and 224 were developed the sheer man hours involved in listening and tweaking must have been boggling.!

Not mind boggling, just dedication to the art. Anything less shows a lack of respect. Today we have examples of mad love available at every level, including free...

...the memories of sounds that once held sway, like apartment lights passed by on the freeway, are life scenes that just begin to register, replaced by two point beams, fading the once best or, yield an ambassador, to a new world held, by those that care and are willing to spell, the towers, and glory, of the past elite, that will be troubled, humbled, and without a seat, no beat, because the music stops, in the time they keep...



-Casey

[dale116dot7]

I've been working on less fun parts of the reverb. Being able to store a user-defined effect, edit its name before storing it, and dumping effects via MIDI to a PC for eventual inclusion into the presets - or just for storage. No new algorithm news yet, I was getting frustrated with the lack of storing carefully created effect settings so I decided to take many days and write all of that code. The most bearish code was writing to the flash. If it's wrong, the processor just resets thanks to the watchdog timer, with little evidence to lead you to the problem. What ends up happening is the flash memory gets taken over by the programming instruction, and if it's wrong, the flash does not return to the CPU and the program just goes bye-bye. Anyways, that's working, including a plea to please name the effect something meaningful. I used a 2x40 display and my effect names are 30 characters long.

Next order of business is to calculate the RT's for display in a more meaningful way. Just trying to get some cleanup done before getting too far ahead of myself.

[zmix]

Originally Posted by Casey

Not mind boggling, just dedication to the art. Anything less shows a lack of respect. Today we have examples of mad love available at every level, including free...-Casey

Preaching to the choir here Casey...
I appreciate the intense amount of work involved in voicing a reverb. I certainly spend weeks listening listening listening then making small tweaks, and more listening listening. I wonder if a publicly held company can afford to postpone a product release until it's 'right'? Perhaps I should post those PCM examples with the modulation OFF so that my point is a bit more obvious. Maybe if I hadn't spent the last several decades "immersed in reverb" I would agree with the reviewers... Barry Blesser describes this phenomenon nicely in an AES paper he sent to me recently:

Barry Blesser:
An artificial reverberation algorithm designed by an
amateur is likely to sound very good to an inexperienced
listener. The author, who has spent hundreds of hours lis-
tening to artificial reverberation algorithms, has infor-
mally observed an improvement in sensitivity to certain
subtle artifacts by at least 20 dB. Informal learning is not
always apparent. Designers of artificial reverberators and
virtual auditory displays must therefore be warned: users
often experience an unintentional learning after extensive
use. By transforming themselves from naive listener to
“golden ears,” they can become intolerant to the “de-
graded”performance. It is hard to predict whether the tol-
erance to imperfections will rise or fall with use, since
both happen.

[dale116dot7]

Originally Posted by zmix

Preaching to the choir here Casey...
I appreciate the intense amount of work involved in voicing a reverb. I certainly spend weeks listening listening listening then making small tweaks, and more listening listening. I wonder if a publicly held company can afford to postpone a product release until it's 'right'?

Thanks to flash memory, that's less of an issue than it used to be. You can send out a product with known deficiencies or maybe just missing that last algorithm that isn't right, then send out an update that users can do with little more than a USB cable or a MIDI adapter.

[seancostello]

Originally Posted by zmix

The modulation artifacts I'm demonstrating here may not be phase shift networks but the tonality is similarly disturbing. I just cannot believe that these algos made it to final release...

My guess is that they are intentional, not accidental. What sounds bad to some ears can sound great to others. The modulation may sound bad on piano, but really nice on drums. Some aspects of the PCM96 may have been half-baked upon release (the Firewire implementation sounds like it had some issues), but the algorithm developer has been developing reverbs at Lexicon for the better part of a few decades at this point, so I am sure that all sonic artifacts are intended.

I will admit that the chorus on the PCM96 Concert Hall sounds very dissimilar to the PCM70 Concert Hall chorus, so I am wondering how accurate of a port there was of the 224 Concert Hall. However, it has been close to 2 decades since I have worked with a 224XL, and I have never seen a 224, so maybe this is what it sounded like.

Sean Costello

[Froombosch]

Originally Posted by zmix

Preaching to the choir here Casey...
I appreciate the intense amount of work involved in voicing a reverb. I certainly spend weeks listening listening listening then making small tweaks, and more listening listening.

So many right words.

[zmix]

I've decided to post another comparison of the PCM96 "Concert Hall" and the PCM-70 (while I am waiting for the 224 to arrive).

This time the reverbs are in < - STEREO - > and you will hear the 'wet' signal only.

In order to hear the tail better, I stretched the decay out to 6.3 seconds. All settings are the same on each unit, no early reflections, however, I found that I needed to increase the size of the PCM96 to 53m in order to match the loop length of the PCM-70 at 33.3m...

In the previous examples you could hear that the PCM96 modulation was quite different from the 224/ PCM70 modulation.

Now you will hear what the algos sound like with no modulation....

[seancostello]

Originally Posted by zmix

Now you will hear what the algos sound like with no modulation....

The algorithms sound more similar without the modulation. They also sound SUPER ringy.

Sean

[zmix]

Originally Posted by seancostello

The algorithms sound more similar without the modulation. They also sound SUPER ringy. Sean

Can you be more specific or elaborate on your observations a bit for us? Also, you mentioned above that: "the algorithm developer has been developing reverbs at Lexicon for the better part of a few decades " who are you referring to?

When I listen to these I do not find these two versions of the same algo all that similar...

Listen to the stereo spread. What accounts for the narrow phasey tail in the PCM96?


As for the actual algo design, I find that the ringing in the PCM-70 is much more benign and shows better eigentone distribution than the PCM96. The PCM96 actually changes the chord voicing of the piano! This characteristic is present in ALL of the PCM96 algos, btw.

Why do we modulate taps and FB in the first place? To tame the runaways! The chorusing in the PCM-70 (and the 224) is magical and lovely. The PCM96 is cyclic, grindy and not very pretty. Combine the wooley ringy algos and unsatisfying modulation and we have... a commercial release from a major brand name?

Is there no 'there' there anymore?

[seancostello]

Originally Posted by zmix

Can you be more specific or elaborate on your observations a bit for us? Also, you mentioned above that: "the algorithm developer has been developing reverbs at Lexicon for the better part of a few decades " who are you referring to?

Nobody Special - the user name on Gearslutz, not a snide answer. Check out his Gearslutz profile. Also, read his posts on the Lexicon bestiary thread:

Lexicon reverbs: a brief bestiary

Listen to the stereo spread. What accounts for the narrow phasey tail in the PCM96?

It sounds wide to me, but I am using headphones. I would think that there might be a difference between the mono summing of the PCM70, and the true stereo of the PCM96 (and the 224).

As for the actual algo design, I find that the ringing in the PCM-70 is much more benign and shows better eigentone distribution than the PCM96. The PCM96 actually changes the chord voicing of the piano!

I just hear discrete echos and metallic allpass ringing in both of the reverbs with the modulation off. It seems like this particular algorithm REQUIRES modulation to sound halfway decent. The PCM70 sounds like it has a bit more high frequency attenuation, but I hear some very pronounced sine waves ringing out in the tail. If I listened closer, I could probably hear more subtle differences, but since both of the algorithms sound bad without chorusing, this is where I will stop.

Why do we modulate taps and FB in the first place? To tame the runaways! The chorusing in the PCM-70 (and the 224) is magical and lovely. The PCM96 is cyclic, grindy and not very pretty.

If the Concert Hall algorithm is as dependent on modulation as your examples would indicate, then the particular type of chorusing is probably critical to the sound.

Sean

[zmix]

Originally Posted by seancostello

Nobody Special - the user name on Gearslutz, not a snide answer. Check out his Gearslutz profile. Also, read his posts on the Lexicon bestiary thread:

Lexicon reverbs: a brief bestiary

Thanks for the reply Sean...

Not taken as snide, Sean..

Is NS really a programmer? I didn't get that impression from what I've read here.

It sounds wide to me, but I am using headphones. I would think that there might be a difference between the mono summing of the PCM70, and the true stereo of the PCM96 (and the 224).

1: you cannot image using headphones.
2: Mono source in both cases, and this shouldn't affect the tail. I am convinced that there is something wrong with the PCM96 routing.

I just hear discrete echos and metallic allpass ringing in both of the reverbs with the modulation off. It seems like this particular algorithm REQUIRES modulation to sound halfway decent. The PCM70 sounds like it has a bit more high frequency attenuation, but I hear some very pronounced sine waves ringing out in the tail. If I listened closer, I could probably hear more subtle differences, but since both of the algorithms sound bad without chorusing, this is where I will stop.

If the Concert Hall algorithm is as dependent on modulation as your examples would indicate, then the particular type of chorusing is probably critical to the sound.
Sean

I'll post more examples of different algos. I cannot afford to 'stop' as you say since I'm trying to incorporate this box into my current mix setup, or bail on it entirely. As it stands I may have to stockpile vintage Lex parts unless I can get to the bottom of this 96 voicing problem.

[seancostello]

Originally Posted by zmix

Thanks for the reply Sean... I'll post more examples of different algos. I cannot afford to 'stop' as you say since I'm trying to incorporate this box into my current mix setup, or bail on it entirely. As it stands I may have to stockpile vintage Lex parts unless I can get to the bottom of this 96 voicing problem.

Stockpiling vintage Lex parts is probably always a good idea.

Not to toot my own horn or anything (and I will now proceed to do so), but have you checked out Eos? Chris Randall is a big fan of the PCM70 Concert Hall algorithm, so I designed something with a similar sonic goal, but with less coloration (even with modulation turned off) and more echo and eigenmode density. It isn't a Lexicon clone, so there are undoubtedly sonic characteristics that are far different from the PCM70, but I think it sounds pretty good. The modulation is different from both of the Concert Hall examples you posted, so it will not directly emulate that part of the PCM70, but I can get similar results with reduced modulation depth and increased modulation speed.

Sean

[zmix]

Sean,
Eos sounds quite interesting, it is certainly very 'pretty' (which is exactly what I was hoping for from Lexicon).

Congrats on a successful product...!

Quite a bold move to leave off the size parameter...why?

[seancostello]

Originally Posted by zmix

Sean,
Eos sounds quite interesting, it is certainly very 'pretty' (which is exactly what I was hoping for from Lexicon).

Congrats on a successful product...!

Quite a bold move to leave off the size parameter...why?

Size is on the top left, second parameter down. Under predelay, over decay.

Sean

[Warp69]

Originally Posted by zmix

2: Mono source in both cases, and this shouldn't affect the tail. I am convinced that there is something wrong with the PCM96 routing.

Do you feed both channels with the same signal on the PCM96?

[zmix]

Originally Posted by Warp69

Do you feed both channels with the same signal on the PCM96?

Yes, a mono source to all reverbs.

[zmix]

Here's an interesting contrast to the previously posted examples:

This is Sean's Eos reverb same settings as above, 33.3m, no modulation, 6.3s Hall algo etc...

[zmix]

Here are a collection of algo based HW and SW plate reverbs, with and without modulation.


all 17.7m, 2.3/3.6 seconds, etc.....

[seancostello]

Originally Posted by zmix

Here are a collection of algo based HW and SW plate reverbs, with and without modulation.

all 17.7m, 2.3/3.6 seconds, etc.....

The mod rate and depth is set kinda high on the Eos examples, IMO. For the plate algorithm, I tend to set the freq around 0.7 Hz or so, and the depth around 1/3 to 1/2. The parameters give the user enough rope to hang themselves, if deliberately chorusy reverbs are desired, but the plate algorithms tend to work with lower amounts of modulation than the hall algorithm, at least when I was working with it. On the other hand, your other sound examples are about turning up the modulation depth and speed high enough so that the modulation is obvious, so in that sense the examples are probably appropriate.

Thanky you for posting these examples, BTW. It is nice to hear the same audio example for all of these platforms. I haven't heard the Sony R7 before - it sounds quite nice.

Speaking of modulated halls, have you tried the Ensoniq DP/2 and DP/4 Large Halls in comparison with the PCM70 and PCM96? It was my understanding that these algorithms shared some DNA with the 224 algorithms.

Sean

[zmix]

Originally Posted by seancostello

The mod rate and depth is set kinda high on the Eos examples, IMO. For the plate algorithm, I tend to set the freq around 0.7 Hz or so, and the depth around 1/3 to 1/2. The parameters give the user enough rope to hang themselves, if deliberately chorusy reverbs are desired, but the plate algorithms tend to work with lower amounts of modulation than the hall algorithm, at least when I was working with it. On the other hand, your other sound examples are about turning up the modulation depth and speed high enough so that the modulation is obvious, so in that sense the examples are probably appropriate.

That was the point of my demonstration. I matched the modulation rate and amount in the PCM96 and the Eos specifically to illustrate the odd placement of the modulation in the PCM.. what were they thinking? I think that you've come very close to some of the 'prettier' reverb sounds I tend to gravitate towards with your Eos, but to my tastes it's the Plate 1 algo that has the best magic.

Thanky you for posting these examples, BTW. It is nice to hear the same audio example for all of these platforms. I haven't heard the Sony R7 before - it sounds quite nice.

I have several early Sony reverbs. The R7 is the only 'modern' one. They are incredible. SONY put - so - much R&D into these algos.. and they are fantastic. Nothing in them is modulated or randomized, they are simply tweaked to perfection.

Speaking of modulated halls, have you tried the Ensoniq DP/2 and DP/4 Large Halls in comparison with the PCM70 and PCM96? It was my understanding that these algorithms shared some DNA with the 224 algorithms.

Sean

As I mentioned before, I worked as a consultant with Ensoniq to develop the DP/4+ , DP/2 and ultimately the DP/Pro. Many of the reverb presets in the DP/2 and DP/Pro are my work. The DP/Pro has the famous "Dattorro algorhythm". I pleaded with Jon to give me access to modulation for some of the taps and we did end up with a very powerful device. I own a DP/2, a DP/4+ and two DP/Pro (including one with a prototype Digital I/O. That piano track I have been using for the examples comes from a record I mixed a few weeks ago and on the actual mix, I used the DP/pro with a factory preset I created called "Black 224".....

I printed the reverb return back to tape, and here it is

Note the low density, high modulation and late 1970s primitivism....

[seancostello]

Originally Posted by zmix

I think that you've come very close to some of the 'prettier' reverb sounds I tend to gravitate towards with your Eos, but to my tastes it's the Plate 1 algo that has the best magic.

That algorithm is less Lexicon, more EMT. It isn't really EMT, any more than Superhall is Lexicon, but it shares some design goals. I've never heard an EMT-250 in person, so my understanding was based on impulse responses and the like.

I have several early Sony reverbs. The R7 is the only 'modern' one. They are incredible. SONY put - so - much R&D into these algos.. and they are fantastic. Nothing in them is modulated or randomized, they are simply tweaked to perfection.

I am reminded of the Bricasti samples / impulses I have heard. Super high echo density (almost like convolution with noise), but with no ringing at all. The decay does not sound entirely exponential to me, which I am sure is by design.

The DP/Pro has the famous "Dattorro algorhythm". I pleaded with Jon to give me access to modulation for some of the taps and we did end up with a very powerful device.

Is this "the" Dattorro algorithm, or "a" Dattorro algorithm? The Halls on the DP/2 I have sound considerably different from the Small and Large Plate algorithms, and much closer to the Lexicon PCM70 Concert Hall algorithm I have heard.

I used the DP/pro with a factory preset I created called "Black 224".....

Which algorithm? One of the Expert Reverbs? Did the older Hall algorithms from the DP/4 and DP/2 make it over to the DP/Pro, or was something else used for the Hall presets?

Note the low density, high modulation and late 1970s primitivism....

Nice. It doesn't sound low density to me at all, but that could be the modulation masking this.

Sean

[dale116dot7]

Tuning is starting to drive me a bit crazy. There are a few things I'm hearing that I really need to understand a bit better how to solve. Allpass ringing. With modulation it smooths out standing wave buildup but there's still a lot of ringing when you hit it with something like a rim shot, unless the gains are set very low. But then the looping tap pattern is obvious. It's a grainy ringy sort of sound - a combination of the loop length and tap pattern (as you'd expect), along with a ringy allpass sound. The tap pattern you can change to randomize the sound, but even modulated allpasses have a periodic 'ringing' sort of sound.

Am I being too hard on myself or is a clean rim shot sound a good tuning tool? I find that running vocals or other instruments through pretty much any reverb seems to sound 'ok' unless you listen for a while to really tune your ears to it, but running a rim shot through it seems to point to faults in the algorithm or tuning right away. I've especially noticed that when I compare a PCM91 to the multi-loop algorithm from earlier, they both sound pretty good on vocals. But the sound is much smoother on the '91 on a rim shot - you can hear individual taps but all of the reflections sound pretty random in time.

I usually use the 'vocal magic' preset on the '91, that's probably my favorite vocal setting unless I'm doing something really dry - then I use some ambience setting, or something really spacey - then I use a PCM90 with a bunch of chorus and reverb and crank it up way too high.

[zmix]

Originally Posted by seancostello

That algorithm is less Lexicon, more EMT. It isn't really EMT, any more than Superhall is Lexicon, but it shares some design goals. I've never heard an EMT-250 in person, so my understanding was based on impulse responses and the like.

I used the EMT250 quite a bit on records in the 1980s and 1990s. The EMT 250 is much more sparse, contains more spacial cues and has a 'constant density-esque' chilling vastness and dare I say "organic" interactiveness with the source material. There is an inexplicable quality to the 250 which I shall attempt to describe in words by saying that is gives back to each source as needed... you can use a single 250 across several disparate instruments and somehow not feel like it's the same reverb on everything. That for me is the unseen genius in the 250. That took years of study and some very insightful design. The impulses I've heard seem to fail in that regard. Did you read this? It's like a game of chinese telephone out there!

I am reminded of the Bricasti samples / impulses I have heard. Super high echo density (almost like convolution with noise), but with no ringing at all. The decay does not sound entirely exponential to me, which I am sure is by design.

White noise convolution is an interesting theory, but if we have learned antthing from the EMT250 / Lexicon 224 and sadly the PCM96, you cannot underestimate the value of human beings voicing reverb by LISTENING.

Is this "the" Dattorro algorithm, or "a" Dattorro algorithm? The Halls on the DP/2 I have sound considerably different from the Small and Large Plate algorithms, and much closer to the Lexicon PCM70 Concert Hall algorithm I have heard.

Which algorithm? One of the Expert Reverbs? Did the older Hall algorithms from the DP/4 and DP/2 make it over to the DP/Pro, or was something else used for the Hall presets?

Nice. It doesn't sound low density to me at all, but that could be the modulation masking this.
Sean

The DP/Pro was purpose built from the ground up using the (then) new "ESP2" chip.

The DP/Pro shares no code with the DP/2 or DP/4 series.
The only 'reverbs' are the two "Expert Reverb" 1 and 2 algos.
These are used for all of the internal reverb presets on the DP/Pro.
There is a 'reflection modeler' and a 'nonlin' as well, but these are not 'reverbs'per se...
The "Dattorro Algo" is THE one in his subsequent papers.
Here is the description from the manual:

ENSONIQ DP/PRO manual
Description: A fully customizable stereo reverberation module. The delays that make up the
reverberator can be adjusted to create a variety of plates and spaces. Features low and high shelving
EQ, a six-stage diffuser, predelay, low and mid decay times, density control, and high-frequency
damping.

Here is the parameter list for the "Expert Reverb" note that the input has 6 diffusors independently adjustable for the left and right inputs, and 4 modulation LFOs

[Warp69]

Here's a test with the Bricasti - with and without modulation. Used the dry mp3's from zmix as sourcefiles.

[zmix]

Originally Posted by Warp69

Here's a test with the Bricasti - with and without modulation. Used the dry mp3's from zmix as sourcefiles.

It is difficult to hear what the reverb is doing, can you print the summed outputs of the reverb to one channel and pan the source to the other?

[zmix]

Ensoniq DP/2 Hall 6.3s


Oddly the Ensoniq DP/4 and DP/4 reverb algos have no size parameter, but two diffusion parameters.

This is the DP/2 "Hall Reverb" algo at 6.3 seconds decay.

[Michael Carnes]

Originally Posted by zmix

Is NS really a programmer? I didn't get that impression from what I've read here.

Hmm. What's a programmer supposed to sound like on a B.B. thread?

zmix, I'm sorry you've taken such a dislike to the PCM96. There are an awful lot of very good people in the industry who've told us it's the best-sounding box we've ever made. I do a bit of classical recording on the side, and I think it's head and shoulders above our earlier boxes for that application.

Since many of the complaints seem to have something to do with modulation, let's talk about it. The purpose of modulation in Lex boxes has never been about being obvious--it's always been about breaking up room modes and improving the spectral performance. In the case of the old ConcertHall algorithm, and to a lesser extent the RandomHall algorithm, this had mixed results. While spectral performance was improved, the modulation was obvious. For a great many users, this turned out to be a plus. They loved the effect. David Griesinger, who wrote those algorithms, was never wild about that effect--especially in ConcertHall. Getting any information from him on exactly how the modulation worked was like pulling teeth. He simply wasn't interested. That's why the 96 sounds a bit different from the older boxes. Unfortunately, the original 224 and PCM70 code was lost long, long ago.

The newer algorithms like Hall and Room have a much more successful modulation, if you consider the original goal. Unless pushed to an extreme (spin anywhere past about 2, 2.5 Hz), you can't really tell it's going on.

So as I said before, sorry you don't like it. But I'll still pop in whenever I can and answer questions and comments.

N.S.

[Casey]

Originally Posted by Nobody Special

David Griesinger, who wrote those algorithms, was never wild about that effect--especially in ConcertHall. Getting any information from him on exactly how the modulation worked was like pulling teeth. He simply wasn't interested. Unfortunately, the original 224 and PCM70 code was lost long, long ago.

David spoke with me many times and in great detail about the concert hall algorithm and the modulation methods it used. PM me.

AFAIK David does still have the 224 code base. It was the 480 code base that was somehow destroyed.

In fact, David has wondered as to whether it would make sense to re-release the 224 as a plugin. Perhaps he and Barry will get together. Barry has mentioned to me that he would like to re-release the 250. We were speculating on another thread that perhaps this is the reason that UAD has been talking up the 250 in their recent news letter.



-Casey