Reverb Subculture

[seancostello]

Of the 4 examples, I like the Sony plate the best. By far. The density of the algorithm is really nice, and I don't hear much metallic coloration. The reverb sounds like a natural extension of the drums.

The other reverbs have too much L-R flutter for my tastes. The PCM96 is OK, but the 224 is WAY too fluttery. The EMT250 sounds grainy to me in this context.

It would be interesting to hear the same settings on pitched material, as what sounds good for drums might sound too metallic on piano.

[seancostello]

The Universal Audio EMT250 was announced today. In the video on the product page, you can see Barry Blesser talking about the original hardware, and that it was intended to emulate a hall. You can also see a blurry drawing of the ALGORITHM about halfway through the video. It looks a lot like what Casey described earlier in this thread: 3 parallel combs, with 3 taps per comb. The taps are combined by something that is obscured by a big graphic of the EMT250 box, and then sent into 3 or 4 delay lines. There is a bunch of other stuff that is hard to read. Still, interesting that they would use the algorithm drawing as part of their promotion. I don't think I would do that for my own work.

Sean

[zmix]

Originally Posted by seancostello

The Universal Audio EMT250 was announced today. In the video on the product page, you can see Barry Blesser talking about the original hardware, and that it was intended to emulate a hall. You can also see a blurry drawing of the ALGORITHM about halfway through the video. It looks a lot like what Casey described earlier in this thread: 3 parallel combs, with 3 taps per comb. The taps are combined by something that is obscured by a big graphic of the EMT250 box, and then sent into 3 or 4 delay lines. There is a bunch of other stuff that is hard to read. Still, interesting that they would use the algorithm drawing as part of their promotion. I don't think I would do that for my own work.

Sean

If anybody wants your algorhythm drawings in 33 years consider it an honor..!

The 250 has 3 delay taps for the left and three for the right, though the 1st tap on the right channel is a straight pass through and has no delay unless the predelay is set to 20, 40 or 60 ms. The rear outputs do not seem to have these same taps but just the combs matrix...

I've got to find and watch that UA video!!

EDIT: HERE it is.....

[seancostello]

Originally Posted by zmix

I've got to find and watch that UA video!!

EDIT: HERE it is.....

Watching in high quality is fun. The drawing is visible at 2:32. Not sure if this is EMT250, though - the date of the drawing is "Blesser Associates, June 77." This might be the EMT250, or it could be a later EMT algorithm, or something else. Still fun to look at.

Sean

[dale116dot7]

Anyone with EMT250 EPROM's???

[Warp69]

Originally Posted by seancostello

Not sure if this is EMT250, though - the date of the drawing is "Blesser Associates, June 77."

It is.

[Froombosch]

Sorry Dale I have not, but we are producing a 250 lookalike for VST in the very near future...

[Froombosch]

Originally Posted by Warp69

It is.

+1 confirmed

[seancostello]

Originally Posted by Froombosch

+1 confirmed

Looks VERY close to description on Froombosch's forum from a few years ago. Not sure what is going on with the left side of the drawing. At 2:36, there is a crossfade where you can see the whole drawing pretty well.

Are we nerds or what?

[audioawake]

if anyone interested in comparision between xVerb and IR(480L):
Free File Hosting Made Simple - MediaFire
Preset was autopark, matched as close as I could get it.
Added a dry file also, so maybe someone with real 480 might jump in.

[Coyoteous]

Originally Posted by seancostello

Are we nerds or what?

Yes, but this is one of the best threads ever because you are!

I have a lot of experience using and tweaking reverbs that I've talked some about in other threads. The closest I've come to programming them is modifying Csound verbs... some of Sean's in fact, IIRC and modifying scripts (or whatever they are called) for Bill Gardner's "Reverb" application... anybody else play with that?

Okay... back to lurking mostly, but thanks to all the thread contributors!

[seancostello]

Originally Posted by Coyoteous

I have a lot of experience using and tweaking reverbs that I've talked some about in other threads. The closest I've come to programming them is modifying Csound verbs... some of Sean's in fact, IIRC and modifying scripts (or whatever they are called) for Bill Gardner's "Reverb" application... anybody else play with that?

Wow, those Csound ones are ANCIENT! The last one I did was an 8-delay FDN (I called it a waveguide reverb, but it wasn't really) back in October 1999. It was kinda sparse for the initial decay. Today I would put 3 or 4 series allpasses in front of it, or after the outputs - the allpasses Casey describes in this thread would do a good job of increasing the stereo image.

I never played with Gardner's Reverb application, and I don't know if I have any machines that would run it today (and wouldn't know where to find an Audiomedia card to run the audio in real time). It was a pretty cool idea, though: two delay lines, with operators that can read and write to and from the buffers. I have kept the program around, and look at the presets from time to time.

Sean Costello

[zmix]

Originally Posted by dale116dot7

Anyone with EMT250 EPROM's???

Have you seen this?

EMT 250 REVERB MANUAL AND FULL SCHEMATICS 76MB Contains 1 PDF file - osCommerce :

[dale116dot7]

That might be useful in combination with the contents of any ROM chips on the boards - just a schematic doesn't necessarily help unless everything is implemented without any programmable memory chips at all.

Reversing anything is a lot of effort that can better be spent learning the design aspects. What I have found is after visiting the code of, say, a PCM70, LXP1, and DN780, you learn less and less from further reversing, and you work on bigger boxes that take longer and longer to figure out.

Sometimes archival reversal is interesting... for example the enterprising individuals that have built copies of the old Apollo guidance computer... running the original code.

[seancostello]

Originally Posted by seancostello

This is assuming that (1-x) is being calculated on the host, and not on the DSP.

Looking at the EMT250 algorithm diagram from the UA video, it looks like the short delay lines that process the summed comb output taps are calculating (1-x) on the DSP itself. The short delay lines have 3 inputs into a summer:

- 2 output taps, each associated with a scale
- the 2nd output tap is also sent to the summer, with no scale

Assuming that the signs are as follows:

- Tap 1 (scaled): +
- Tap 2 (scaled): -
- Tap 2 (unscaled): +

then the linear interpolation can be performed on the DSP, using a single interpolation coefficient.

The standard linear interpolation equation is:

out = frac*read[tap] + (1-frac)*read[tap+1]

The EMT250 diagram looks like it is calculating

out = frac*read[tap] + read[tap+1] - frac*read[tap+1]

If multiplies are more expensive than additions & subtractions on a given processor, this could be calculated as

out = frac*(read[tap]-read[tap+1]) + read[tap+1]

All of these will probably have different advantages or disadvantages, depending on your processor, numeric format, etc.

[seancostello]

For anyone who has read this far in this thread, you might be interested in my blog:

The Halls of Valhalla

I have been writing about general reverb issues, as well as some more techie stuff. Not nearly as techie as on this thread, but it might be of interest to y'all. My latest writings have been about modulation (definitely inspired by zmix's sound files), as well as general ways of working with the controls in reverb algorithms. I also talk about the "lost" Schroeder algorithm in an earlier post.

Feel free to comment on the blog if you are so inclined. I'll be hanging around this thread as well, flipping my virtual nickel in the air and chewing on my virtual toothpick.

[zmix]

Originally Posted by seancostello

For anyone who has read this far in this thread, you might be interested in my blog:

The Halls of Valhalla

I have been writing about general reverb issues, as well as some more techie stuff. Not nearly as techie as on this thread, but it might be of interest to y'all. My latest writings have been about modulation (definitely inspired by zmix's sound files), as well as general ways of working with the controls in reverb algorithms. I also talk about the "lost" Schroeder algorithm in an earlier post.

Feel free to comment on the blog if you are so inclined. I'll be hanging around this thread as well, flipping my virtual nickel in the air and chewing on my virtual toothpick.

Well, I have a few bones to pick, Sean... first of all Jeff Buckley, not a Quadraverb. There is a stereo flange/ chorus from an SPX-90 added in mix and a Lexicon 224xl.

Secondly "First Light" on the Eno / Budd record is certainly a 250, but it's in conjunction with several passes of 250 (using the 'chorus" program on a printed reverb pass, for example). Thw 250 in and of itself is not nearly that lush...plus it has only a 4.5 second decay (it seems less) sne there is a fair amount of digital delay augmenting that track.

[zmix]

When speaking in terms of modulation and 'naturalness' I think that there is a arbitrary and needlessly distancing cognitive split between reverb 'purists' and the dismissal of 'effect designers'.

Even the UAD 250 video has Dr Blesser lamenting the (implied) descent into 'the pure effect world' instead of the simulation of pure acoustic phenomenon.

Sony, for example, refused to add any type of modulation to their first and second generation reverberators, stating in the inimitably corporate way, that these things "did not exist in nature".
In a similar vein, I have posted some fairly descriptive examples in this thread of modulation gone wrong in a modern state of the art reverberator, and the response from the programmer fell pretty much on the side of the 'purist', even stating that in 'classical music' these things do not apply.

But is this "purist" vs. "effect" philosophy valid?

Dr Blesser mentioned in his 2001 AES presentation that there is a considerable amount of pitch and amplitude randomness in natural acoustic spaces due to thermal currents and other forms of air movement.

I say that we drop the often condescending split between 'purist' and 'effect' approach and try to understand WHY these modulations are so important to reverb design, beyond 'simulating greater modal density' and 'eliminating ringing'.

I recently read a paper by David Griesinger in which he described the inability to discern exact pitch as an aural cue for the perception of distance. Clearly Mr. Griesinger is an exceptional thinker and as a reverb designer remains on the vanguard to this day.

I highly recommend this paper as it provides yet another vital piece to the puzzle......

[seancostello]

Originally Posted by zmix

Well, I have a few bones to pick, Sean... first of all Jeff Buckley, not a Quadraverb. There is a stereo flange/ chorus from an SPX-90 added in mix and a Lexicon 224xl.

Secondly "First Light" on the Eno / Budd record is certainly a 250, but it's in conjunction with several passes of 250 (using the 'chorus" program on a printed reverb pass, for example). Thw 250 in and of itself is not nearly that lush...plus it has only a 4.5 second decay (it seems less) sne there is a fair amount of digital delay augmenting that track.

Would you mind putting the above in comments in the blog?

Also, I can get some identical sounds to Buckley's guitar sound out of my noisy Quadraverb, using the CHORUS REVERB program. I had read about the Quadraverb being used by Buckley. Having said that, I am by no means a Jeff Buckley obsessive - he's OK. I was just looking for modulated reverb examples. Where did you hear about the SPX-90/224xl combo?

As far as multiple passes of the 250 on the Budd/Eno stuff, that makes sense. I figured that there was some feedback delay, but I was having difficulty gauging the amount of chorusing found in the EMT250 reverb algorithm, and the modulation rate used in the reverb, since I haven't used an EMT250 in person.

The last song on the album, "Falling Light," is more obviously chorused, so I figured that the EMT250 Chorus algorithm was being used. It sounds really nice.

So, what does the EMT250 Space program do?

[zmix]

Originally Posted by seancostello

Where did you hear about the SPX-90/224xl combo?

It's what Andy Wallace used on the record.

Originally Posted by seancostello

So, what does the EMT250 Space program do?

Oddly, it's an unmodulated looping not very wide 'reverb effect' that I don't think anyone ever used on a commercially successful record....

[seancostello]

Originally Posted by zmix

When speaking in terms of modulation and 'naturalness' I think that there is a arbitrary and needlessly distancing cognitive split between reverb 'purists' and the dismissal of 'effect designers'.

I'm an "effect designer." I wear my badge with pride! I don't come from any sort of classical background. I grew up on pop, punk, ambient, and the soundtrack of "Cosmos."

Dr Blesser mentioned in his 2001 AES presentation that there is a considerable amount of pitch and amplitude randomness in natural acoustic spaces due to thermal currents and other forms of air movement.

Yet I have heard many examples of artificial reverbs where the amount of pitch variation seems to go way beyond what you would hear in a "healthy" space. Your average concert hall doesn't have bonfires in it, which is what would be required for the pitch variations that some reverbs can get.

Having said that, maybe the excess pitch variation is stimulating some cognitive mechanism in us, such that we like it. Time to crack open some psychoacoustics literature.

I say that we drop the often condescending split between 'purist' and 'effect' approach and try to understand WHY these modulations are so important to reverb design, beyond 'simulating greater modal density' and 'eliminating ringing'.

Well, they do that. And they do simulate thermal currents. The question is, how is this important from a perceptual basis?

I highly recommend this paper as it provides yet another vital piece to the puzzle......

I'll check it out.

From my point of view, modulated reverbs are a useful tool in taking simple sounds, and making them sound far grander. A simple synth or sustained guitar note, through a well-modulated reverb, takes on a complexity of tone that is almost orchestral. The micro-fluctuations in pitch, multiplied by feedback, create a sound that is like a hugely complicated ensemble chorus.

I have toyed around with algorithms that generate a reverberant impression, but without actually sustaining the sound like real acoustic reverberation. Phase randomization, plus lots of subtle parallel pitch modulations, seems to be the key. Gotta drag those out again.

[seancostello]

Originally Posted by zmix

It's what Andy Wallace used on the record.

Well, yeah, but besides that...

Oddly, it's an unmodulated looping not very wide 'reverb effect' that I don't think anyone ever used on a commercially successful record....

I wonder why the modulation was turned off, instead of down?

[zmix]

Sean,

The main point I was trying to make concerning re-evaluating modulation as more than just 'affect' is contained in the statement I made (above) about David Griesinger's paper:

"...the inability to discern exact pitch as an aural cue for the perception of distance..."

Though that statement is the logical converse of his study, it is a potent idea, and I believed it was true intuitively after reading that paper...

[Casey]

FZ fan much zmix?

I thought I was the cool kid on the block with all this reverb tech, until a woman just down the street converted light into matter and then back again. No shit!

Man, I really need to up my game.



-Casey

[EgyptionRing]

Originally Posted by zmix

We have the EMT 250, the original Lexicon 224, the Lexicon PCM96 and the Audio Damage Eos here... care to guess?

Well, I would like to know because all I can afford is the Eos. Any hints?

[zmix]

Originally Posted by Casey

FZ fan much zmix?

I thought I was the cool kid on the block with all this reverb tech, until a woman just down the street converted light into matter and then back again. No shit!

Man, I really need to up my game.



-Casey

You are what you is, my friend.....

[Casey]

Originally Posted by zmix

You are what you is, my friend.....

Now THAT is the crux of the biscuit!



-Casey

[zmix]

Originally Posted by Casey

... a woman just down the street converted light into matter and then back again..
-Casey

Wow, if the inverse were possible.... beam me up Scottie!

[seancostello]

Originally Posted by zmix

The main point I was trying to make concerning re-evaluating modulation as more than just 'affect' is contained in the statement I made (above) about David Griesinger's paper:

"...the inability to discern exact pitch as an aural cue for the perception of distance..."

Though that statement is the logical converse of his study, it is a potent idea, and I believed it was true intuitively after reading that paper...

The inability to discern exact pitch that Griesinger describes only occurs with a select group of signals, primary solo vocalists. This is due to the impulsive nature of vocal waveforms - even though the vocal tract adds resonances, the primary nature of the sound is impulsive, and the instantaneous amplitude of the signal varies according to the fundamental frequency of the voice. A small amount of decorrelation of the vocal signal results in the periodic variation of the instantaneous amplitude being lost. It is interesting that we perceive this as reverberation, or distance.

Other sustained musical signals do not have such correlated harmonics, and are not perceived as reverberant with the type of decorrelation Griesinger performs. Impulsive signals, such as percussion, rely on the harmonics being in phase sync in order to retain the shape of the impulse, so they will also bring out the perception of distance more than other signals.

[zmix]

Originally Posted by seancostello

The inability to discern exact pitch that Griesinger describes only occurs with a select group of signals, primary solo vocalists. This is due to the impulsive nature of vocal waveforms - even though the vocal tract adds resonances, the primary nature of the sound is impulsive, and the instantaneous amplitude of the signal varies according to the fundamental frequency of the voice. A small amount of decorrelation of the vocal signal results in the periodic variation of the instantaneous amplitude being lost. It is interesting that we perceive this as reverberation, or distance.

Other sustained musical signals do not have such correlated harmonics, and are not perceived as reverberant with the type of decorrelation Griesinger performs. Impulsive signals, such as percussion, rely on the harmonics being in phase sync in order to retain the shape of the impulse, so they will also bring out the perception of distance more than other signals.

Be that as it may, it's certain that random modal decorrelaton adds a sense of depth.

Today, while mixing a rock tune with a close mic'ed string quartet, I decided to try a few experiments with my 224. I was using a fantastic program from the rev 4 software called "ROOM - A" (program 14). It's a very nice stereo algo . I set up a sound on it that really added life to the strings, then I turned off the modulation ...oh, there's the rear wall, there's the ceiling, etc... turning ON the modulation was like "...OH, listen to the rosin, ... nice, here's where the cello takes the melody" etc... NOT "wow man check out the chorusy strings" at all.... In fact, with the modulation off, it sounded nearly identical to my Sony / Ibanez verbs, which is to say great, but turning on the modulation made the walls recede while still providing the sense of the strings 'moving air'... perfect.