The No.1 Website for Pro Audio
 All  This Thread  Reviews  Gear Database  Synths for sale     Latest  Trending
U-He Diva vs Boog Model D: Oscilloscope Waveforms
Old 20th January 2019
  #1
Gear Nut
 
StrangeSatellite's Avatar
 

U-He Diva vs Boog Model D: Oscilloscope Waveforms

So tonight I was A/Bing my Boog Model D with the Minimoog configuration of U-He Diva. For some sounds, the plugin can get quite close to the real thing. For other sounds (plucky sounds with a sharp attack), the plugin doesn't quite capture the same vibe as the real thing, IMO.

So I thought I would take a more scientific look at why this might be by comparing their waveforms. I used only one oscillator with the filter wide open to make this as simple as possible. I'm posting the images in this thread, but if you want to see the full resolution images for more clarity, go here and then click the images to expand them:

Diva vs Model D Waveforms - Album on Imgur

First up, the saw wave. Diva on top, Model D on the bottom:



The thing that really sticks out is that the little wavy parts of the shape where there are sharp changes in the waveform are different. These are called "Gibbs Phenomenon" and you can read more about them here:

Gibbs phenomenon - Wikipedia

In the case of the saw, the Gibbs Phenomenon appears at the top and the bottom of the waveform on the Model D, but Diva only produces it at the top of the saw. It's completely missing from the bottom. The shape of this area at the top is also different, as the Model D seems to have a small dip at the beginning of the Gibbs Phenomenon, where Diva lacks this feature.

Next, square:



Again, same thing as the saw. The ripple in the upper right of the square that you see on the Model D is missing from the Diva waveform.

Last, a triangle:



As you can see, the Model D produces a sharp triangle while Diva shows a bit of ripple.

The point of this thread isn't to start a war of software vs hardware. It's meant to start a discussion as to why these discrepancies exist and how much influence they have on the overall sound. It's interesting that another U-He plugin (Repro) also is missing the Gibbs Phenomenon in the same places. I'm wondering if it's particularly hard to mathematically model for some reason, or whether it was just left out of the emulation for some other reason. Thoughts?
Old 20th January 2019
  #2
Lives for gear
 
eXode's Avatar
 

Quote:
Originally Posted by StrangeSatellite View Post
The point of this thread isn't to start a war of software vs hardware. It's meant to start a discussion as to why these discrepancies exist and how much influence they have on the overall sound. It's interesting that another U-He plugin (Repro) also is missing the Gibbs Phenomenon in the same places. I'm wondering if it's particularly hard to mathematically model for some reason, or whether it was just left out of the emulation for some other reason. Thoughts?
My thoughts is that these type of comparisons are kind of meaningless, and here's why:

First of all, you haven't compared Diva to the specific hardware specimen that it was modeled from.

When it comes to analog in general, and the Minimoog in particular, it's a well known fact that there are discrepancies even between hardware units, so how can you, or anyone, rule out the possibility that the differences you see (or rather, HEAR) are software vs hardware related, when they might just as well be sample related (sample as in specimen)? This is the problem with pretty much all comparisons between emulative software and hardware.

I think the best approach is to view it from the whole rather than getting into semantics. Does it sound something like a Minimoog in a convincing manner, rather than, does it sound exactly like my Minimoog from 197X with serial 12345 and the early revision oscillator board?

That's my 2 cents anyway.
Old 20th January 2019
  #3
Lives for gear
 
BobTheDog's Avatar
 

What makes you think these "Wavy Parts" are being created by Diva or the Borg?
Old 20th January 2019
  #4
Lives for gear
 
stinkyfingers's Avatar
 

Any filter in the signal path will change the shape of a waveform. (Duh...)

So you post a link about Gibbs and then wonder why that waveform looks like that ? Are you serious?

Last edited by stinkyfingers; 20th January 2019 at 01:01 PM..
Old 20th January 2019
  #5
Lives for gear
 
enossified's Avatar
I recorded the sound of my snowblower and checked out the waveforms on a scope.The Gibbs phemomenon looked wrong, so I sold it and bought a different brand.
Old 20th January 2019
  #6
Lives for gear
 

I agree that on this era of synths (and their current condition) it makes sense that there could be a wider range of experiences and thus which specific synth you modeled may lead to this.

However, I think the OPs interest is valid. One possible hypothesis on the results:

Maybe U-he (or whomever) only models the Gibbs effect at the start of the waveform and not at the conclusion.

You could further speculate this could be for a few reasons:
1. It was determined that the Gibbs effect at end of waveform had less of an effect on tone
2. Processing limitations
3. Something else

I think it’s cool to think about and discuss this
Old 20th January 2019
  #7
interesting tests, although to be honest Im not a huge fan of raw oscillator tests...they are really the ground level aspects of an emulation that we should expect to be very very close as to be essentially audibly indistinguishable. Not to mention - they are boring musically.

When it comes to sound I am more interested. I think I know what you are talking about re the pluckiness of the envelopes and it is something I also have noticed. However, I challenge you to capture this as audio, level match and present it here. You would be amazed at how quickly differences disappear when presented blind.
Old 20th January 2019
  #8
Lives for gear
 
monomer's Avatar
 

Quote:
Originally Posted by StrangeSatellite View Post
The ripple in the upper right of the square that you see on the Model D is missing from the Diva waveform.
None of those ripples are actually missing.
They are just different in phase and in the case of diva cancel out on one end.
So what you're looking at is basically a phase shift in these high frequencies.

Another thing is that with the boog these ripples are actually caused by your audio interface (assuming it's indeed gibbs phenomenon). They occur when a signal is sharply band limited. The boog itself doesn't do that and any audio interface does do it.
So these ringing things are basically a consequence of the sampling system and not a real indicator of the true waveform that was sampled.

The triangle looks a bit different, but again, this can be due to phase shifts and if that is true would have no effect on how it sounds.

Last edited by monomer; 20th January 2019 at 06:11 PM..
Old 20th January 2019
  #9
Lives for gear
 
monomer's Avatar
 

Quote:
Originally Posted by BobTheDog View Post
What makes you think these "Wavy Parts" are being created by Diva or the Borg?
In case of Diva they are created by Diva's algorithms.
In case of Boog they are created by the audio interface.
Old 20th January 2019
  #10
Lives for gear
 
monomer's Avatar
 

Quote:
Originally Posted by Jean Luc Cougar View Post
However, I think the OPs interest is valid.
Why do you think this?
Specifically, what relation do you think there is between a waveform on a scope and how you hear that waveform?
Old 20th January 2019
  #11
Lives for gear
 
Shadowkast's Avatar
 

It might be interesting to see a comparison of current soft synths and VA's and those of say, 10-15 years ago. Compare all these different results and we can see where modelling started, and how far it's come to this day in a visual sense.
Old 20th January 2019
  #12
Lives for gear
 

I'm curious to see a scope comparison between Diva and the Model D, but with those sounds that you thought were very similar, and then one with those plucky sounds with sharp attacks where they sounded different to you, instead of just the raw waveforms unfiltered etc. With text explaining mixer levels, filter settings, envelope, and so on. Then maybe we could figure out what contributes most to the differences between the software model and your specific model D. By a process of elimination, if nothing else.

From what you've said so far, it sounds like the attack portions of their envelopes aren't identical. On the oscillator scopes, you've shown that the triangles are the least similar, but still very, and also that they're all phenomenal, with the real Model D being slightly more phenomenal.
Old 20th January 2019
  #13
Gear Maniac
 
ac_music's Avatar
 

How about we make music and stfu with all this analysis of waveforms. I'm sure some of you buy gear just because you analyze every detail and waveform like some kind of autistic lunatic. For the newbs reading: it doesn't matter what gear you use, only the end result matters. And the end result can be achieved with any gear, many many different ways. I can use one instrument VST and a couple of effects and produce albums for 1000 years - the gear matters only so much as it's core functionality and basic features.
Old 20th January 2019
  #14
Lives for gear
 
stinkyfingers's Avatar
 


https://i.imgur.com/wqCxtRH.mp4

(sorry about the quality, i'm not good at video)
Old 20th January 2019
  #15
Lives for gear
 
BobTheDog's Avatar
 

Quote:
Originally Posted by monomer View Post
In case of Diva they are created by Diva's algorithms.
In case of Boog they are created by the audio interface.
Or the scope
Old 20th January 2019
  #16
Lives for gear
 

Quote:
Originally Posted by monomer View Post
Why do you think this?
Specifically, what relation do you think there is between a waveform on a scope and how you hear that waveform?
First and foremost - I’m no expert, I’m simply curious. I’m way too lazy to do this sort of thing, but I enjoy reading, learning and hypothesizing when I can’t make music.

The relationship in my mind is the scope is a visual representation of what you hear.

The OP heard something different and decided to try to confirm what he was hearing. He visually saw differences.
Old 20th January 2019
  #17
Lives for gear
 
monomer's Avatar
 

Quote:
Originally Posted by BobTheDog View Post
Or the scope
Not likely. The scope is operating at the samplerate already. It would have to interpolate unnecessarily to get any sort of extra wiggles.
But this is maybe something that needs to be cleared up.
Old 20th January 2019
  #18
Lives for gear
 

measure times the damping factor in the bass, yes know difficult .. but that is a big difference ..
a simple method is tube amp and bass speaker .. because you see the difference quite well ... va makes at the same level much more travel
Old 20th January 2019
  #19
Lives for gear
 
monomer's Avatar
 

Quote:
Originally Posted by Jean Luc Cougar View Post
The relationship in my mind is the scope is a visual representation of what you hear.
Aah, well, this is a false assumption.

A spectrum analyzer would be much closer to what we actually hear.
A waveform is not very indicative of how something sounds, most of the time.
I can produce waveforms that all sound like squares, saws and triangles but which nevertheless don't look anything like squares, saws or triangles.
So it's a bit difficult to judge things.

When you see a saw wave it will very probably sound like a saw wave.
But when you see a different kind of wave (and not square or triangle) you wouldn't be able to tell if it sounded like a saw wave as well.

So this makes 'waveform spotting' a bit problematic.
Old 20th January 2019
  #20
Gear Guru
 
zerocrossing's Avatar
I think you are missing the point of Diva. Urs has said on many occasions that he changed things when he felt he liked the results more than a more close emulation of the original hardware. He also took license to change things to better obtain 16 voices of polyphony at a reasonable CPU load. (Try that with your Boog!) Use osc three in audio rate mode to modulate the filter in Diva and then try it on your Boog. Diva sounds kind of dull. I wish that Urs would add an “ultra-divine” mode for times when you’d compromise voices for better audio rate modulation. However, if you look at his RePro 1/5, he nailed it.

Now, if you want to compare apples to digital apples, check out Legend. That gets a lot closer to a Model D, and I bet you hear less difference.
Old 20th January 2019
  #21
Lives for gear
 
jupiter8's Avatar
 

DIVA is modeled on a Minimoog Hans Zimmer thought sounded particulary good. I recall another MM A/B test and DIVA sounded quite different from the others. My guess is it is because of the particular MM it models, it simply sounds like that.
Old 20th January 2019
  #22
Lives for gear
 
monomer's Avatar
 

Quote:
Originally Posted by Jean Luc Cougar View Post
The OP heard something different and decided to try to confirm what he was hearing. He visually saw differences.
What is a possible cause for a different sound is the artifact in the Boogs waveform.
Notice how, around the big changes in amplitude, the Boog raises the waveform a little. I think this is switching noise caused by imperfections of the oscillator. A VCO has some 'mechanics' to it. Things need to flip, currents need to be shunted to different parts, that kind of thing. And it takes some time because contrary to what almost everyone think, analog is not infinitely fast.
And THAT is something that will for sure give an extra 'bite' to the boogs sound. Rough around the edges as they say.
Apparently Diva doesn't emulate this kind of switching imperfections.

Would be interesting to see if a moog does this as well and if so, how much of it you can see.
Old 20th January 2019
  #23
Gear Nut
 
StrangeSatellite's Avatar
 

Quote:
Originally Posted by ac_music View Post
How about we make music and stfu with all this analysis of waveforms. I'm sure some of you buy gear just because you analyze every detail and waveform like some kind of autistic lunatic. For the newbs reading: it doesn't matter what gear you use, only the end result matters. And the end result can be achieved with any gear, many many different ways. I can use one instrument VST and a couple of effects and produce albums for 1000 years - the gear matters only so much as it's core functionality and basic features.
I make plenty music. Also, like many people here, I enjoy the technical aspect of synthesis. This isn’t a songwriting forum, after all.
Old 20th January 2019
  #24
Gear Nut
 
StrangeSatellite's Avatar
 

Quote:
Originally Posted by monomer View Post
In case of Diva they are created by Diva's algorithms.
In case of Boog they are created by the audio interface.
Interesting. So if I were to use an "analog" scope, I wouldn't see the wavy endpoints?
Old 20th January 2019
  #25
Lives for gear
 
EvilDragon's Avatar
You would, because again, audio interface would create them on the outputs (antialiasing filters in the DAC).
Old 20th January 2019
  #26
Lives for gear
 
stinkyfingers's Avatar
 

Quote:
Originally Posted by monomer View Post
Notice how, around the big changes in amplitude, the Boog raises the waveform a little.
are you talking about the little "kink" just after min/max ?
that's there because the ADC is using an elliptical or chebyschev filter...
Old 20th January 2019
  #27
Lives for gear
 
monomer's Avatar
 

Quote:
Originally Posted by StrangeSatellite View Post
Interesting. So if I were to use an "analog" scope, I wouldn't see the wavy endpoints?
In case of the boog directly on the scope you will probably not see these wiggles. But there may be some other wiggles much higher in the spectrum.

In case of diva you would see wiggels and they would be related to the samplerate (like in the case of sampling the boog).
Old 20th January 2019
  #28
Lives for gear
 
monomer's Avatar
 

Quote:
Originally Posted by stinkyfingers View Post
are you talking about the little "kink" just after min/max ?
that's there because the ADC is using an elliptical or chebyschev filter...
What i see is a kindof square-ish offset starting just before a transition and ending just after the transition.
Not sure i've seen this happen with chebychev filters but i don't know a lot about elliptical filters.
Maybe you're right and its caused by some interesting interactions of phase.
Old 20th January 2019
  #29
Gear Nut
 
StrangeSatellite's Avatar
 

Quote:
Originally Posted by EvilDragon View Post
You would, because again, audio interface would create them on the outputs (antialiasing filters in the DAC).
Right, but what about an analog scope that doesn't have a DAC?
Old 20th January 2019
  #30
Lives for gear
 
BobTheDog's Avatar
 

Quote:
Originally Posted by monomer View Post
Not likely. The scope is operating at the samplerate already. It would have to interpolate unnecessarily to get any sort of extra wiggles.
But this is maybe something that needs to be cleared up.
Ah you must know what scope the measurements were made on then?
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump