U-No-LX VS Diva

[SadPuppyBlues]

Wait a second.

There are VAs that can compete with DIVA now?

No way!
When did that happen?
What are they?

[urosh]

It is very (very!) likely that TAL is cooking up a SH101 clone as well, so with a little more patience we'll get even more close!

great, now I'll have to wait.

There are VAs that can compete with DIVA now?

There are VAs that were and still are better than Diva for one or another thing. I spend quite some time trying to match some Oxium basses with Diva and she was utter failure (not only that she would not match exactly, but sounded worse for same kind of patches). And, I think LuSH101 does better job with glassy mid80s rolandish pads than Diva, although I think it's failure in nailing sound of SH101. Not to mention that Diva will never pull off some stuff my MS2000 does. But all in all, as far as analog emulation goes, Diva is reference point ATM.[/quote]

[Uncle E]

I've been testing U-NO-LX for last 3 days and Diva is my main ITB synth. I've tried to match them as close as posible and you can get very very close but there is still some minute differences that can be important for nitpickers like me.

Excellent review!

Is it worth $40 if you already have Diva?

$40 doesn't buy much, these days. To have a plug-in that directly replaces a $600-800 piece of hardware cost so little is a beautiful thing.

IMHO, when you count in different workflow (U-NO is simply dead easy to program, cant get simpler than that) and lower CPU load it is worth another $40 to add it to Diva.

Is the CPU load that much lower? Have they made it more efficient recently? When I tested it, the CPU loads weren't much different.

[fmr]

Is the CPU load that much lower? Have they made it more efficient recently? When I tested it, the CPU loads weren't much different.

I don't know if it's "much lower" (how much is much?) but definitely "lower" (quite lower) than DIVA.

[Examigan]

Wait a second.

There are VAs that can compete with DIVA now?

Yes, I get along just fine without it

[Uncle E]

Yes, I get along just fine without it

Actually, no, you're missing out.

[urosh]

Is the CPU load that much lower? Have they made it more efficient recently? When I tested it, the CPU loads weren't much different.

It is on my system, about 30% lower with mono patches and lastest Diva build. Bigger issue is that CPU load is pretty much constant with U-NO and Diva load dynamicaly depends on patch and UI contols.

[BDeep]

There was a version that had a significantly higher CPU usage, that was an error and quickly resolved by TAL. I believe this was only on the Mac version though. Anyway, on my system, yes,U-NO-LX is significantly lighter on the CPU than Diva.

[Uncle E]

Great! I'm on a Mac and it may be that I was using that higher CPU build. Thanks, guys!

[PAK]

BTW, I've matched only 3 patches. Right out of the box U-NO feels much more different, so maybe I just haven found areas where they differ more.

Yep, there's more. I think Uno LX does several things better, in terms of Juno sounds. But I suppose you'd hope so when it's a dedicated emu

Listen to the polyphony handling on longer release sounds.. it's quite surprising what a difference that makes. There's something strange about the Poly2 mode in Diva btw, but I've never taken the time to figure what. The subtle voice differences (of the 60 hardware) are more closely approximated in Uno LX too. In Diva you sometimes have to assign an LFO to the DCO, in addition to the individual voice detuning under trimmers, in order to approximate things more closely (and I still think Uno LX manages more of a "vintage hardware" vibe anyway, even after doing stuff like that to Diva).

Above 50% I think Diva's chorus can behave more like a send effect, rather than integrating with the sound, so that can make for some quite noticeable differences (generally in UNo LX's favour) where sounds have a stronger emphasis on the chorus.

Get into the subtleties of the filters resonance (which is easier from a controller than the GUI) and Uno LX can easily pull out some nice psuedo-reverb sounds which I struggled (and failed) to get from Diva's DCO model through the cascade filters. This is, again, more true to Roland hardware.

Perhaps the biggest difference, though, is the envelopes. Uno LX's behave significantly more like the hardware to me, except in one aspect. In common with many emulations it lacks some perception of punch in the attack stage, which will be noticeable for things like fast arp sounds and stuff like synth bass/brass etc. Diva seems to use some tricks here, so it fairs better in preserving those qualities of analogue. But otherwise you might find yourself tweaking a lot more as Diva seems to like sticking you in narrow filter envelope mod sweet spots, and I often find myself multiplying or adding the envs in Diva to get an envelope shape I'm happier with too.

It's several months since I spent a few hours comparing things, so hopefully my memories held up I like to keep the number of plugins I use to a minimum, so that's why I bothered. Uno LX easily passed my expectations, and I think it's fair to call it the best native plugin emulation (of a specific analogue synth) out there.

Diva's more versatile, of course. So it's not quite an Apples to Apples comparison. But I think, if you're even slightly interested in the sound of vintage hardware, Uno LX is very worthwhile.. Still wouldn't mind seeing a more flexible 2 DCO version though

[urosh]

In Diva you sometimes have to assign an LFO to the DCO, in addition to the individual voice detuning under trimmers, in order to approximate things more closely (and I still think Uno LX manages more of a "vintage hardware" vibe anyway, even after doing stuff like that to Diva).

I agree. Thing is, I've discussed this with Urs, and I think it's philosophical issue. He really doesn't like to do things outside of the users control and sight (you'll remember that he considers secret EQing under the hood as first class felony) and he likes his synths to be repeatable and "well mannered". And it's perfectly valid and understandable standpoint.
BTW, I don't get far with just LFOs, way to static/predictable, I sometimes use this trick: I set LFOs to modulate each others frequency, and such cross coupled cicular modualtion arangements are one of the paths to realm of chaos. However, I use up both LFOs and most of CV processors where one pink noise source with strong 1/f component per oscilator could do wonders. And I think Diva wouldn't mind at all if some kind of Wogglebug mode for LFOs would be added.

But otherwise you might find yourself tweaking a lot more as Diva seems to like sticking you in narrow filter envelope mod sweet spots, and I often find myself multiplying or adding the envs in Diva to get an envelope shape I'm happier with too.

Exactly the same situation here, 90% of tiresom tweaking with Diva is trying to get envelopes the way I want them (and in half of situations I simply can't quite get there). BTW, perception of smoother attack in U-NO is not as bad, some of analogs I like (and here it's very much the case of personal preferences) have "smoother" midrange (I know, descriptions of tone sucks) compared to VAs and dedicated emus, and to my ears it's most evident when faster attacks sweep filter trough midrange, so less "chewy" env in U-NO woruks to his advantage in some occasions (BTW, there is also an issue of VCF response to CV change; ladders are uberfast and cap-to-ground OTA VCFs in roland gear have a bit laggy response that is dynamicaly dependant on cutoff freq; thankfully, those popular topology preserving approaches to DSP filters preserve these properties as well even if developer is completely unaware of that issue)

[PAK]

(you'll remember that he considers secret EQing under the hood as first class felony)

Well, I understand that one. Having to EQ means your models are falling down somewhere, and those failures will still be evident to discerning ears no matter how much you attempt to mask them. Better to try to fix something than use a crutch to hide issues.

Kind of ironic, though, that there was talk of an optional muffle parameter being added for upper frequency differences in Diva when the likely reasons for the differences (in the perception of "smoothness" esp of higher frequencies of analogue hardware Vs digital) is likely to be a far more fundamental and complex modeling problem than merely dampening frequencies to mask differences. Yep, that will also give a less harsh result, but it's likely to also be a big over-simplification, which will be evident in the results

so less "chewy" env in U-NO woruks to his advantage in some occasions

Mmm, chewy. I get what you mean though and, yes, differences like that can sometimes help too. It's one of those things which most modelled software seems to fall down with though. We need someone to come along with a paper that will do the same for env attack as what happened for filter resonance

[urosh]

Kind of ironic, though, that there was talk of an optional muffle parameter being added for upper frequency differences in Diva when the likely reasons for the differences (in the perception of "smoothness" esp of higher frequencies of analogue hardware Vs digital) is likely to be a far more fundamental and complex modeling problem than merely dampening frequencies to mask differences.

Main problem is that problem is not well defined. When you do something like emulation of filter you mostly work by specs, you have to meet that and this well defined criteria, same thing as with design of low alias oscillators. But this issue we're discussing is whole 'nother ball game. I mean, I say "Divas midrange sounds hard to me compared to vintage SEM" but what the heck "hard", "harsh" or "soft" mean? And it's not analog modeling problem per se, some very analog stuff sounds very hard to my ears (like Polivoks VCF for example, about which OTOH half of modular comunity raves about), but still, main issue is very poor definition of underlying problem and lack of deterministic well behaved metric to deterimine differences.

And, same story applys to this

We need someone to come along with a paper that will do the same for env attack as what happened for filter resonance

as well. Present 10 different synths to 10 different KVRians and you'll likely get different answers which one is "punchy" (or whatever). And without definition of "punch" or "silky high frequencies" we'll not see any scientific work on those issues (or "issues" for some people).

[fluffy_little_something]

I like to stack an instance of the LX and an instance of Charlatan together, and add a separate LX chorus to that combination, that yields quite a full, flexible quality sound as both sound clean and compliment each other well, with Charlatan adding the sizzle to the solid LX foundation

[Sendy]

Wasn't the "muffle" idea for Diva just lowering the upper frequency range of the filters to more resemble the operating range of the original gear, and not eq-ing out high frequencies after the fact?