Diva vs Analogue - a real world test

[kvaca]

I don't need scientific isolation, whatever that is. It would be enough if the patches were as close as possible, which is not the case with some sounds. Test 5 for instance, they could have been made to sound much closer. The obvious difference there is not due to hw vs sw, but patch design.

agreed, exactly!!!

also - test 10 is completely off which prevents any reasonable comparation on sound level
also - test 9 has major difference in vibrato fadein
also - test 5 has hearable distortion on B examples only
also - test 2 is completely off /much faster hifreq.rolloff on exaple B/

[Hank the Knife]

'??
what do you call the whole zoo in this test?
do you mean theres something important missing?'
@kvaca
I meant that I did hear other instruments too on the given examples. So it's the opposite: there's too much going around in these tracks to compare the synths in a strict sense ONLY, I suppose. Thought I made that clear. Well.. somehow. Maybe this helps.
@fluffy
The patches should be equal. If that's possible.

[fluffy_little_something]

Were the drum sounds also made with the Oberheim and Diva?

[Hank the Knife]

Concluding: the presentation of two synth which sounds are 'matched' -tweaked if you like- in advance to make them sound 'equal' doesn't say anything about any comparison at all between the two synths. On the contrary I'd say. It just says that the poster of these vids tried to make them SOUND equal and then asks us to determine which is which. Yeah right.. I think that is a misunderstanding as it comes to a reasonable approach when comparing the SOUND of the synths in question. Hey.. I've got other things to do!

[kvaca]

true, but the one sounding best had to be the hardware for many

Thats the depressing truth. Many people still assume that analogue synths have the superior sound in tests like these. Many people still want the software to fail, to confirm their bias.

You forgot that similarly many biased people want software to win, isnt it strange?
Or its normal that some happy owners of Diva /or OB/ want their instrument to win the battle in any case?

I couldn't care less which side wins, but what I really hate is snobbery in all its forms (which you find anywhere in this world, seems to be a human constant in a way...)

yes, exactly...its a human feature, one among others...you can always hate it, and thats all you can do with it

[Gamma-UT]

If we use the narcissism of small differences in this thread to judge, software and hardware are pretty much on an equal level now.

When people start arguing the toss over whether a high-frequency rolloff is symptomatic of a particular sound generator or whether a bit of constant-level noise added to foil the visual sample inspectors is a problem, it's getting very close indeed. And it's slightly baffling to me given that there are other tells between the two that mean you don't really have to use those differences of patch design to make a fairly consistent distinction between the two.

Synth A generally has a higher level of subtle HF ('air' for want of a better word) probably from saturation/distortion that comes across to my mind as a little natural ambience. It's a quality that's common to a lot of analogue hardware (and it's the thing that Bob Carver added to his cheap audiophile-confounding amp designs to make them sound more audiophiley). If that is Diva, then hats off to both the synth and the patch design. But I'd vote A as the hardware being the more likely culprit.

I'd be perfectly happy to use B and add that 'air' myself if that's what the sound design needs. It's likely to get buried under even the softest reverb.

[Urs]

The Diva (some 2 years old demo, didn't bother to install the latest one, sorry) was running at the "Great" rather than "Divine" quality, but that was already about the double CPU usage of Monark (the host SR was set to 88kHz).

I thought about that. Hehehe, I guess I need to fire up Monark again. So if Monark is faster than Diva on extreme settings, it should either alias more, or it uses a limited number of iterations to converge, or it uses an analytical approach. When I checked Monark out a few months ago I had the impression that it generally used more CPU than Diva for a single voice, but I haven't tried extremes. And it might have been a biased impression.

Monark is always running at 88/96kHz regardless of the host SR. OTOH, I had impression from Diva, that the internal sampling rates are configured relative to the host. Is that true? Because then maybe "Great" at 88kHz host SR is already "Divine" at 44?

As a general remark, it would be nice if developers took into account that people use other host SR's than 44 and 48. I was interested in Lush-101 once, but it was unusable CPU-wise for me at 88kHz. My guess is that it oversamples internally by the same factor, no matter what the host SR is.

Well, filters in both Great and Divine mode are commonly oversampled 4x, relative to host sample rate. Divine has smaller margin error for convergence than Great, hence takes more iterations to finish at settings with high resonance.

We've discussed internal sample rate vs. host sample rate a few times. Our stance is that people expect better quality when they go 88.2/96 kHz. Which they wouldn't get if we didn't oversample by the same factor internally - we would actually need to downsample to 44.1/48khz to achieve the same level of bandlimiting.

Maybe one day we'll add it as a preference.

(I think hosts should let users choose different sample rates for different tracks... whatever is required for the task at hand)

Edit: BTW, if you ever get to do a completely revised version of ACE, I'd like to be a part of the beta (not sure, if you would like me to )

Sure, I don't see why not unless your contract with your employer stands against such thing.

[Hank the Knife]

'I couldn't care less which side wins, but...'
@27rgE
That's like saying (analogy): 'I am not a racist, but...'
Speaking of snobbery..

[SWAN808]

Answers:

Synth A: u-he Diva with Classic OB presets



http://swanaudio.co.uk/

Synth B: Oberheim OB8




Hope you guys enjoyed the test...I will leave up the wav downloads for as long as possible...

[SWAN808]

PS aside from the added noise (recorded from the OB8 and kept around typical analogue noisefloor of -60db), there was no unfair shennanigans in this test. I set out to do it as fairly as I could. Diva was recorded live at Great mode, only 1 cut EQ was used on both instruments - otherwise it was just levels and a shared reverb bus. (+plugin masterbus compression and EQ shared). If its not clear - the Song Test was 1 project file with each instrument playing 4 bars each then switch.

[Z1202]

As a general remark, it would be nice if developers took into account that people use other host SR's than 44 and 48. I was interested in Lush-101 once, but it was unusable CPU-wise for me at 88kHz. My guess is that it oversamples internally by the same factor, no matter what the host SR is.

Well, filters in both Great and Divine mode are commonly oversampled 4x, relative to host sample rate. Divine has smaller margin error for convergence than Great, hence takes more iterations to finish at settings with high resonance.

We've discussed internal sample rate vs. host sample rate a few times. Our stance is that people expect better quality when they go 88.2/96 kHz. Which they wouldn't get if we didn't oversample by the same factor internally - we would actually need to downsample to 44.1/48khz to achieve the same level of bandlimiting.

Would you care to reveal some of the internal quality setting details regarding Diva? Basically my impression is, that the settings in Diva are configured to be most reasonably usable at 44kHz host SR, correct? So, if I have to work at 88kHz host SR instead, what are the "nearest match" Diva quality settings to those at 44kHz host SR?

[chk071]

PS aside from the added noise (recorded from the OB8 and kept around typical analogue noisefloor of -60db), there was no unfair shennanigans in this test. I set out to do it as fairly as I could. Diva was recorded live at Great mode, only 1 cut EQ was used on both instruments - otherwise it was just levels and a shared reverb bus.

Hm, is Diva really so much brighter than OB-8? And does the OB-8 really have such a muffled, odd low end, and odd oscillator sound to it? Hard to believe tbh... well i don't know what you did with the EQ so. Not playing bad loser here, but i guess myself, and the lot of other people who voted incorrectly kind of could have sworn that it would be the other way around. At least i was 99% sure.

Well, i guess what has been said in the other thread is true then, that Diva was made to have a brighter top end, because that's the one thing i noticed first when comparing the high end, that it's a lot more present, and wider in example A.

[Ingonator]

Answers:

Synth A: u-he Diva with Classic OB presets



http://swanaudio.co.uk/

Synth B: Oberheim OB8




Hope you guys enjoyed the test...I will leave up the wav downloads for as long as possible...

Wow, so i and many others were wrong with synth A being the real OB8. Good job of both you and Diva.

[wagtunes]

Well, even though I did think the two were very close and couldn't hear a big difference, I did say that I liked the bottom end of Synth B a little more. I thought it was a little richer and fuller and that turned out to be the hardware synth. So maybe hardware still does have a slight edge in sound quality.

Regardless, the sound quality is close enough for me that I'll still use a soft synth over a hardware synth for the following reasons.

Cheaper
Weightless
Takes up no space
No cables needed
Less electricity used
No repairs ever
Can have as many instances as my PC can handle

Those reasons alone are enough for me to choose software over hardware 100% of the time unless I have to play out live which I never do as I am a home studio musician.

Your mileage may vary depending on your needs.

[Lotuzia]

Answers:

Synth A: u-he Diva with Classic OB presets
....
Synth B: Oberheim OB8

Hope you guys enjoyed the test...I will leave up the wav downloads for as long as possible...

Thanks for make me win D.S OB-8 on the other forum

And for the game.

Results as expected. What I would have liked to have more fun are compared poll results in the Sw oriented forums, where most people are already half convinced of one side of the things, and on the HW oriented forum, where most people are already half convinced of the opposite. THAT would have been fun

+ Nice job on the presets. True, obvious differences there are, but you made them probably the less obvious possible.