The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Synths for sale  Search Gearslutz Go Advanced
Analogue vs Digital - Diva and OB8 test Keyboard Synthesizers
View Poll Results: Which synth is First in the 8 bar cycle?
Diva
92 Votes - 51.11%
OB8
88 Votes - 48.89%
Voters: 180. You may not vote on this poll

Old 25th July 2014
  #901
Urs
Lives for gear
Quote:
Originally Posted by Autumn Leaves View Post
"Making a software emulation more reliable and stable than a vintage synth makes it, well, more reliable and stable, but it also keeps the give aways intact, such as chorus noise and stuttering PWM."

So did you keep the chorus noise intact in the software emulation, which would contradict your first statement? Or is there an option to choose between original chorus and noise-free chorus?
I meant: Not having chorus noise is a give away.

That is, the chorus noise of the Juno 60 sucks. Hence we didn't do it. But that of course makes a recording of a Juno 60 easily identifyable vs. a recording of Diva.
Old 25th July 2014
  #902
Registered User
 

Quote:
Originally Posted by Urs View Post
I meant: Not having chorus noise is a give away.

That is, the chorus noise of the Juno 60 sucks. Hence we didn't do it. But that of course makes a recording of a Juno 60 easily identifyable vs. a recording of Diva.
Thanks for your response!

Could you please also answer my question concerning "stuttering PWM"?
Old 25th July 2014
  #903
Lives for gear
 
maisonvague's Avatar
 

Quote:
Originally Posted by Musicncars View Post
Since, the end had come, I will take the corpse off your hands, so you can move on to a completely digital musical realm. :0)
Sorry, but the Prophet has requested that he be cremated and his ashes cast into the audience at the next Genesis concert.

Quote:
Originally Posted by pendejo View Post
Beautiful maison! Out with the old and in with the new...
Thanks, but I'm thinking more like... out with the old and in with the even older!
Old 25th July 2014
  #904
he did.
Old 25th July 2014
  #905
Lives for gear
I would cry if the swishing noise of my Juno 60 chorus disappeared. Its part of its personality.
2
Share
Old 25th July 2014
  #906
Lives for gear
 

Quote:
Originally Posted by Lune View Post
I actually think I know where you are coming from and can help you. You are operating from some premises that are restricting your thinking. When we "know" something it is not always like an on/off switch- which is the assumption you are working with.

Let me give you an example:

If I am presented with my Wife's signature and a child's imitation of it I feel very certain (100% know) which is the real thing. Now, if I see my Wife's signature and a very good forgery I may still feel I know but my certainty is less (maybe 8% know for sure). Now if my wife's signature is shown compared to a copy made by a MASTER forgery artist my certainty may completely disappear (0%).

knowledge is not an all or nothing thing- I doubt anybody who voted- even the experts- would have felt comfortable taking a large bet that they were right. This is what a master copy does to our sense of certainty. I do not believe anybody who voted felt 100% certain- but I could be wrong (and they could have been wrong).

In a sense- everybody was operating with little certainty- this logically leads to a random result.

Does this help?
I understand that, but by that rationale , what is the point of doing these tests?
If there is no 100 percent certainty? Although, I think you are wrong in saying there could be no 100 percent certainty, but I didn't take a statistics class.
Old 25th July 2014
  #907
Urs
Lives for gear
Quote:
Originally Posted by Musicncars View Post
Then why are we even having this discussion, if Diva is not meant to be indistinguishable from analog gear? I thought Urs a stated earlier, that they are trying to sound like or better than analog?
Gosh. It's a more differntiated concept than what you're putting in my mouth.

The question is "does Diva hold a candle against analogue synths". In case of the sound example the OP chose, I guess it's pretty evident. In other cases maybe not, who knows.

However, a completely different question is "does Diva sound better than the analogue synthesizers that guys at u-he had for reference". In this case, and this is by all means a subjective questions, I'd say yes, because we tweaked the algorithms that way. We left undesired properties out, and we added a tad more brilliance here, a tad more bass there and a tad more "warmth" wherever else.

This means, Diva is not supposed to sound "better than analogue". She's just supposed to mimic certain - desirable - aspects of analogue very well, while not sporting aspects that we didn't like.

As an example: When you press a second key on a Minimoog, the pitch goes up by, hmmm, 10 cents or so. It would have been a piece of cake for us to model. But we didn't do so because we think of this as a bug. Therefore we didn't do it.
Old 25th July 2014
  #908
Lives for gear
 
bil_g's Avatar
 

Quote:
Originally Posted by maisonvague View Post
Face it. It's over. Digital has won.


...Fare thee well, my friend!

Reminds me of Oxygene 1.
1
Share
Old 25th July 2014
  #909
Lives for gear
 
grumphh's Avatar
 

Quote:
Originally Posted by Musicncars View Post
what is the point of doing these tests?
The point of doing these tests is to see whether the numbers end up at a different result than 50/50.

If the numbers had been 30/70 in favour of the correct answer Urs would have some long nights of programming ahead of him and everyone could still go on and say that digital can't emulate analogue.

As it is now, you can't, and Urs can take a week off, knowing that some of the worlds finest connoisseurs of analogue synths (by their own admission) can be tricked by his software.
Old 25th July 2014
  #910
Urs
Lives for gear
Quote:
Originally Posted by Autumn Leaves View Post
Thanks for your response!

Could you please also answer my question concerning "stuttering PWM"?
I meant (and I'm sorry for the repeat): Not having a stuttering PWM is a give away.

That is, the stuttering PWM of the OB-8 - as heard in the example - sucks. Hence we didn't do it. But that of course makes a recording of a OB-8 easily identifyable vs. a recording of Diva.

(I have no idea why the PWM sounds like it does in the audio example, but as it was mentioned several times before, I don't think we would have modeled this, had we had an OB-8 as reference. Our OB-Xes sure don't stutter that way)
1
Share
Old 25th July 2014
  #911
Lives for gear
 
Eigenwert's Avatar
The only way to really tell something is to repeat such a test with the same participants and different examples. What if 80% couldn't tell, 20% perceived a difference, with half of the 20% perceiving the DIVA sound as "more in your face", thus analogue and the other half perceiving a, say, somewhat less defined sound as the OB8. The result would still be the same. There are way to many variables in here. The result of the test only tells us something about the crowd, but the 50/50 split does not say that there was no one who could tell, as opposing trends could cancel each other out.

If you ask 10.000 soccer fans whether they want club A or club B to win and 5.000 answer: A; the other 5.000 answer: B

The result is a 50:50 split

now does it automatically mean that all of the fans, including hooligans, can't tell which club they want to win? No, possibly there just were two opposing trends. Thus a 50:50 split does not say anything at all about the individual, if not a whole bunch of things difficult to rule out has been ruled out.
Old 25th July 2014
  #912
Lives for gear
 
grumphh's Avatar
 

Quote:
Originally Posted by Eigenwert View Post
The only way to really tell something is to repeat such a test with the same participants and different examples. What if 80% couldn't tell, 20% perceived a difference, with half of the 20% perceiving the DIVA sound as "more in your face", thus analogue and the other half perceiving a, say, somewhat less defined sound as the OB8. The result would still be the same. There are way to many variables in here. The result of the test only tells us something about the crowd, but the 50/50 split does not say that there was no one who could tell, as opposing trends could cancel each other out.

If you ask 10.000 soccer fans whether they want club A or club B to win and 5.000 answer: A; the other 5.000 answer: B

The result is a 50:50 split

now does it automatically mean that all of the fans, including hooligans, can't tell which club they want to win? No, possibly there just were two opposing trends. Thus a 50:50 split does not say anything at all about the individual, if not a whole bunch of things difficult to rule out has been ruled out.
Another one who could easily be responsible for the early retirement of more than one science/math teacher... heh

Again, this is not a test about hearing differences - this is a test that asks us to attribute the heard differences to analogue or digital sound.
And that is where people failed.
Old 25th July 2014
  #913
Lives for gear
 

Quote:
Originally Posted by Urs View Post
I meant (and I'm sorry for the repeat): Not having a stuttering PWM is a give away.

That is, the stuttering PWM of the OB-8 - as heard in the example - sucks. Hence we didn't do it. But that of course makes a recording of a OB-8 easily identifyable vs. a recording of Diva.

(I have no idea why the PWM sounds like it does in the audio example, but as it was mentioned several times before, I don't think we would have modeled this, had we had an OB-8 as reference. Our OB-Xes sure don't stutter that way)
If it were easily identifyable, more people would have guessed right, no?
Old 25th July 2014
  #914
Urs
Lives for gear
Quote:
Originally Posted by grumphh View Post
The point of doing these tests is to see whether the numbers end up at a different result than 50/50.

If the numbers had been 30/70 in favour of the correct answer Urs would have some long nights of programming ahead of him and everyone could still go on and say that digital can't emulate analogue.

As it is now, you can't, and Urs can take a week off, knowing that some of the worlds finest connoisseurs of analogue synths (by their own admission) can be tricked by his software.
Well, I'm happy, but I also need to be realistic. If it took SWAN longer to match Diva to the OB-8 than, say, a Prophet 5 rev 3 would have taken, then it really is more about sound design skills than quality of emulation.

We've seen threads on KVR where people only *pretended* they used Diva, when in fact they used a decent freeware synth with extraordinarily good programming.

The aim of Diva however is to bahave like analogue gear. That is, it should feel analogue when you tweak it. Turning the cutoff knob from one frequency to another should have the same outcome in Diva as in the analogue counterparts. One should not need to then adjust envelopes, EQ, volume, resonance or anything else to hear the same sound.

This is what it is about. Given enough time, enough sound design skills, an EQ, a saturation plug-in and a decently equipped soft synth, anyone can recreate a good many of classic analogue sounds within a margin narrow enough to fool most people. Our claim with Diva however is, one can do so in less time and without cheapting (eq/saturation) because she behaves right out of the box.
1
Share
Old 25th July 2014
  #915
Lives for gear
 
Eigenwert's Avatar
Quote:
Originally Posted by grumphh View Post
Another one who could easily be responsible for the early retirement of more than one science/math teacher... heh

Again, this is not a test about hearing differences - this is a test that asks us to attribute the heard differences to analogue or digital sound.
And that is where people failed.
There is not even such a thing like analogue or digital sound. The test asked us to identify DIVA vs OB8, how is that not a test about hearing differences.

The way of testing only works for well-defined questions. Define what a analogue or digital was of sounding is. Otherwise you are very likely to get close to a 50:50 result, as one might think "brighter = more digital" and the other might think "brighter = more defined = better = analogue".

The less well-defined the question, the more likely the chance to be close to a 50:50 result as there is one to understand the question one way for anyone thinking something completely different had been meant.
1
Share
Old 25th July 2014
  #916
Urs
Lives for gear
Quote:
Originally Posted by Musicncars View Post
If it were easily identifyable, more people would have guessed right, no?
Well, many people talked about the PWM issue. Just, many people thought it was shortcommings of the softsynth rather than the analogue one and I guess thus they voted wrong.
Old 25th July 2014
  #917
Registered User
Quote:
Originally Posted by m127f View Post
OK, so, I don't know how long Swan spent on matching the sounds, but I am positive Diva does not sound "better" than a real OB8.
Now I've tried them both.
It doesn't sound better than a real OB8, or it doesn't sound the same as a real OB8?

Personally, I don't get the bit about the assumption that "Diva doesn't sound the same as an OB8, therefore the OB8 is better."

Also, this was just a couple of sounds for a few bars. I cannot believe the amount of debate and discussion of minutiae that has resulted from this extremely narrow and limited "test."
1
Share
Old 25th July 2014
  #918
Lives for gear
 
Eigenwert's Avatar
Urs, that's interesting, because if "many" people voted wrong due to one phenomenon an equal amount of people might have voted right due to (an)other phenomenon(s) (say "less brightness"). One gets this all the time with such tests if hidden variables are involved.

BTW I would expect that at least one "expert" really could tell: Urs
since DIVA isn't even modeled directly after the OB8

this thesis stands against the idea that really no one could tell (what the 50:50 result would suggest in case all hidden variables [e.g. one thinks "OB8=dull" and the other "OB8=$$$expensive$$$=in your face=bright"] had been ruled out)
1
Share
Old 26th July 2014
  #919
Lives for gear
 
grumphh's Avatar
 

Quote:
Originally Posted by Eigenwert View Post
There is not even such a thing like analogue or digital sound. The test asked us to identify DIVA vs OB8, how is that not a test about hearing differences.

The way of testing only works for well-defined questions. Define what a analogue or digital was of sounding is. Otherwise you are very likely to get close to a 50:50 result, as one might think "brighter = more digital" and the other might think "brighter = more defined = better = analogue".

The less well-defined the question, the more likely the chance to be close to a 50:50 result as there is one to understand the question one way for anyone thinking something completely different had been meant.
...so now the test was flawed?

The question "is Diva or the OB-8 responsible for the first 4 bars in a piece of music" was to undefined?

Yeah, i can totally see that
Old 26th July 2014
  #920
Registered User
 

Quote:
Originally Posted by Urs View Post
I meant (and I'm sorry for the repeat): Not having a stuttering PWM is a give away.

That is, the stuttering PWM of the OB-8 - as heard in the example - sucks. Hence we didn't do it. But that of course makes a recording of a OB-8 easily identifyable vs. a recording of Diva.

(I have no idea why the PWM sounds like it does in the audio example, but as it was mentioned several times before, I don't think we would have modeled this, had we had an OB-8 as reference. Our OB-Xes sure don't stutter that way)
Thanks again for your answer!

It's funny, because in this test, the stuttering PWM of the analog OB8 sounded like digital processing glitches to my ears, which is why I preferred the other synth, which turned out to be the digital Diva.

It's also interesting in this context, to read the following statement by synth connoisseur and outstanding GS moderator "Don Solaris", whose experienced "analog" ears obviously still preferred the stuttering PWM behavior of the original OB8:

Quote:
Originally Posted by Don Solaris View Post
First synth sounded better. Second synth had slightly better PWM at the point where phases went thru zeros.
Old 26th July 2014
  #921
Lives for gear
 

Quote:
Originally Posted by grumphh View Post
...so now the test was flawed?

The question "is Diva or the OB-8 responsible for the first 4 bars in a piece of music" was to undefined?

Yeah, i can totally see that
By your own admission the test was flawed, doomed from the start.
Old 26th July 2014
  #922
Lives for gear
 

Quote:
Originally Posted by Autumn Leaves View Post
Thanks again for your answer!

It's funny, because in this test, the stuttering PWM of the analog OB8 sounded like digital processing glitches to my ears, which is why I preferred the other synth, which turned out to be the digital Diva.

It's also interesting in this context, to the read the following statement by synth connoisseur and outstanding GS moderator "Don Solaris", whose experienced "analog" ears obviously still preferred the stuttering PWM behavior of the original OB8:



The glitchiness is why I picked diva first, analog is not perfect, and has little random inaccuracies, that is the part of the charm of analog.
Old 26th July 2014
  #923
227861
Guest
we need more tests
Old 26th July 2014
  #924
Here for the gear
 

Quote:
Originally Posted by Eigenwert View Post
Urs, that's interesting, because if "many" people voted wrong due to one phenomenon an equal amount of people might have voted right due to (an)other phenomenon(s) (say "less brightness"). One gets this all the time with such tests if hidden variables are involved.

BTW I would expect that at least one "expert" really could tell: Urs
since DIVA isn't even modeled directly after the OB8

this thesis stands against the idea that really no one could tell (what the 50:50 result would suggest in case all hidden variables [e.g. one thinks "OB8=dull" and the other "OB8=$$$expensive$$$=in your face=bright"] had been ruled out)

i don't think you understand statistics and outliers. if urs can tell that is one person that is statistically insignificant.

also how statistic tests are done you could use another example. Could people tell the difference btwn a 80 dollar merlot made in bordeaux or a 20 dollar merlot made in napa. 52 percent pick the bordeaux 48 pick the napa. yes people might be able to tell there is a differences in the wines but the differences are obviously slight. A Master Somm might be able to tell the difference yes. But considering there are only 219 in the world they would only slightly move the statistical needle.

it's a simple question which synth is which.
Old 26th July 2014
  #925
Registered User
 

Quote:
Originally Posted by Musicncars View Post
The glitchiness is why I picked diva first, analog is not perfect, and has little random inaccuracies, that is the part of the charm of analog.
Of course, but it can also lose its charm, because if you listen closely, the OB8 lead sound begins to badly deteriorate and stutter between 00:42 and 00:46 (and again between 00:57 and 1:01), caused by the PW modulation phases going thru zero, which sounds completely unmusical, while disturbing the whole musical context during those passages... at least to my own ears.
Old 26th July 2014
  #926
Lives for gear
 
grumphh's Avatar
 

Quote:
Originally Posted by Musicncars View Post
By your own admission the test was flawed, doomed from the start.
You really need to quote the post in which i have said that
Old 26th July 2014
  #927
Lives for gear
 

Quote:
Originally Posted by grumphh View Post
You really need to quote the post in which i have said that
Are you serious? The whole time you have said this proves nothing, except no one was right. Pure luck. Therefore, no useable outcome = flawed test. :0)
Old 26th July 2014
  #928
Lives for gear
 

Quote:
Originally Posted by Autumn Leaves View Post
Of course, but it can also lose its charm, because if you listen closely, the OB8 lead sound begins to badly deteriorate and stutter between 00:42 and 00:46 (and again between 00:57 and 1:01), caused by the PW modulation phases going thru zero, which sounds completely unmusical, while disturbing the whole musical context during those passages... at least to my own ears.
It doesn't matter if charm is lost, it was a difference that I noticed, hence I chose diva as the first synth, because of the flaw of the OB8. The test was which synth is first. I listened, and made the appropriate choice, and was correct. So, now, you guys know, at least one of us didn't flip a coin and guess. :0)
Old 26th July 2014
  #929
wow what a thread
Old 26th July 2014
  #930
Registered User
100% Diva. No external effects.

I just recorded this for fun, so take it for what it is.

I like the squishy sound of this. Can an OB8 be even squishier?

2
Share
Topic:
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump