A new competitor to Uhe Diva has arived!!!

[olikana]

monark has the overdrive feedback trick of the minmoog for great distorted tones....Diva does not.
that's the only thing that might lure me to it.
i'd love if Diva had something similar....it would help also in the ms20 department (for the typical phonoout/audioin overdriven ms20 sound).

[Urs]

monark has the overdrive feedback trick of the minmoog for great distorted tones....Diva does not.

What? Diva has a feedback knob. Which does exactly what our Minimoog does when you patch the high out back into the input.

[aboukirev]

In the end we however did not use direct measurements of electronic parts, such as the resistance or cpacitance of the analogue parts. To the contrary, we used our ears.

- Urs

Ahh, so Diva is only as accurate as your (u-he collective) ears are

[olikana]

monark has the overdrive feedback trick of the minmoog for great distorted tones....Diva does not.

What? Diva has a feedback knob. Which does exactly what our Minimoog does when you patch the high out back into the input.

ok...but there's no "load" knob like in the monark to overdrive the signal dramatically . or am i missing something?

[Urs]

monark has the overdrive feedback trick of the minmoog for great distorted tones....Diva does not.

What? Diva has a feedback knob. Which does exactly what our Minimoog does when you patch the high out back into the input.

ok...but there's no "load" knob like in the monark to overdrive the signal dramatically . or am i missing something?

Well, there's no load knob on the Minimoog either... there's just a knob for the external input gain, and that's that.

So, whatever the load knob does in Monark, it isn't anything a Minimoog would do

[Urs]

In the end we however did not use direct measurements of electronic parts, such as the resistance or cpacitance of the analogue parts. To the contrary, we used our ears.

- Urs

Ahh, so Diva is only as accurate as your (u-he collective) ears are

okok... we used a bunch of technical means as well. Analysers, oscilloscopes...

We also used our in-house realtime 4x downsampler so that Howie and I could properly hear what's going on beyond 16kHz

[jalcide]

And what makes you think that circuit modeling itself does not use DSP?

Because of the name, itself: "DSP" (Digital Signal Processing). DSP needs code/computation engine, data memory, program memory and input/output functions to connect with the outside world.

Analog circuits need and use none of that. They use purely electrical components (and not for computation code execution in the digital sense): transistors, capacitors, resistors, etc. -- it's all about the flow of voltages.

They're two very different approaches. Right?

Found this interesting. It goes into both approaches:
http://www.soundonsound.com/sos/aug10/a ... lugins.htm

[jalcide]

Diva's modules are foremost built upon observation of the behaviour of their analogue counterpart.

We built rough digital models based on circuits, but we added some 5-10 additional resistors and current sources, which we could then tweak with knobs. Then we spent months of tweaking those knobs until we felt that we were close. At some point inbetween we might have added more bits an pieces, or we might have tried an alternative for a selected part.

In the end we however did not use direct measurements of electronic parts, such as the resistance or cpacitance of the analogue parts. To the contrary, we used our ears.

- Urs

Interesting. Thanks for the reply, Urs.

Yes, I assumed ears were an important part of the process.

So, even though you may not have followed the analog schematics 100%, that is, you added components as needed to get the desired result, were those added components circuit simulations? Or, is it DSP-based? Or, a hybrid of DSP and circuit simulations?

I will totally understand if you don't want to reveal this, but am very curious where and when physical modeling comes into play (if at all) and when DSP approaches take over. E.g., a high-quality oversampling button would engage a DSP-based set of functions (I would assume). But what about, say, the filter, is it entirely simulating "voltages" for something like that (before it gets sent to another part of the pipeline)?

In short, are there little pockets of circuits being simulated when I run, Diva, for example? Similar to what Softube does with their SPICE stuff? And what Cytomic does with The Drop, Fxpansion with DCAM, etc. I'm assuming Waves also uses this approach with their "Virtual Voltage" thing for Element.

Incredible products. Thanks for everything you do.

Cheers.

[JoeCat]

I guess there's nothing much more flattering than NI producing a Minimoog emulation, and the question on everyone's mind is "how does it compare to Diva?".

Congrats, Urs. You're now in the position of defending the benchmark

[Urs]

So, even though you may not have followed the analog schematics 100%, that is, you added components as needed to get the desired result, were those added components circuit simulations? Or, is it DSP-based? Or, a hybrid of DSP and circuit simulations?

I will totally understand if you don't want to reveal this, but am very curious where and when physical modeling comes into play (if at all) and when DSP approaches take over. E.g., a high-quality oversampling button would engage a DSP-based set of functions (I would assume). But what about, say, the filter, is it entirely simulating "voltages" for something like that (before it gets sent to another part of the pipeline)?

In short, are there little pockets of circuits being simulated when I run, Diva, for example? Similar to what Softube does with their SPICE stuff? And what Cytomic does with The Drop, Fxpansion with DCAM, etc. I'm assuming Waves also uses this approach with their "Virtual Voltage" thing for Element.

I'm not a linguist, but I suspect that "circuit modeling of audio circuitry" is a subclass of digital signal processing.

I personally have no strict definition of either. I just do what I think works best and I don't care what it's called as long as it does what's desired.

What sets The Drop and Diva apart is that both model non-linear aspects of those circuits without commonly used shortcuts. Such shortcuts typically include limitations of the non-linear behaviour (e.g. only 1 simple waveshaper in the filter path) and using fast integration methods. Cytomic and us use different methods to do a similar thing, and despite efforts to optimise the code, they're both computationally more expensive than most stuff. The result speaks for itself.

- Urs

[jalcide]

I'm not a linguist, but I suspect that "circuit modeling of audio circuitry" is a subclass of digital signal processing.

Oh right, absolutely. From an industry category standpoint, or a college, circuit-modeling is probably to be found in a "DSP" course. But, in the spirit of what I'm trying to understand, one *approach* simulates, part-for-part, the behavior of voltages and current in a circuit-block, the other *approach* is more black-box and only concerned with the result. At the end of the day, it's all DSP. I get that.

I just do what I think works best and I don't care what it's called as long as it does what's desired.

Totally agree. As long as all or most of the subtleties of the non-linearities are being captured, who cares how it got there.

The result speaks for itself.

Indeed they do.

All that said, I guess what I may be getting at is that we are still, as of 2013, at a place where there is the potential for improvement to these analog simulations -- great as they do sound today.

To quote from that Sound On Sound article, Wave Arts' Bill Gardner said:

"...the expense grows as the cube of the number of circuit nodes. For example, our Tube Saturator plug-in has six circuit nodes in the non-linear portion, containing two 12AX7 pre-amp stages. This [requires] about 33 percent of a 2.8GHz P4, processing mono 44.1kHz samples; doubling the size of the circuit requires eight times the CPU. A complete tube amp, like a Fender Champ with 24 circuit nodes, would require 64 times the CPU — or about 20 CPU cores! However, CPU power will continue to increase. In the not-so-distant future we'll be able to model any vintage analogue gear by simply entering its circuit schematic and running it in real time. And, in many ways, these simulations will be better than the original circuit: exact component values, perfectly matched components, no noise..."

Right now, it seems that DSP tweaking is about on par with circuit modelling after that circuit model is optimized and dumbed-down to fit into today's CPUs.

One can imagine a not-too-distant future where it's "cheaper" to just dial in the schematics -- no compromises. All those "parasitic effects" of electrical characteristics get solved for free. It just executes those components as if it were real. If the output doesn't match it's not the circuit that's wrong, it's the simulation of it. Once perfected, it would even account for aged components, different brands, damage, etc.

It could make black-box DSP development a lost art. Maybe not, and not for a looooong while.

I'm not unhappy with the current line-up of VA synths (Diva being one of my favorite), but at the same time when I see posts like "oh great, another VA synth emulation" I have to think that these are folks that don't own actual, analog hardware -- cuz there is still room for improvement. Nothing that keeps us from making music, but let's just say that we should all be potential customers of new products, cuz they keep getting better.

[Urs]

Well, at some point the difference becomes rather esoteric.

People believe to hear different things when the user interface is blue or orange, unless they're told that it's the same.

I think that "dumbing down" a circuit model isn't necessarily a bad thing. There's no rule that makes the difference within those lower 5-or-so out of 32 bits sound worse. It may even sound better

Thing is, for future modules in Diva we might need to exaggerate certain aspects of circuits in order to mark clear differences. As in, we'll specifically seek for vintage units that sound a little more colourful than they commonly do. It'll still be true to the concept of emulation, but it'll be in favour of a wider palette of sound.

That said, I'm curious about Monark. The demos seem way more aggressive than Diva. If we can get hold of a Minimoog that sounds *that* aggressive, then we'll happily spend some time on an alternative model for Diva as well. Uhm, if this is generally considered a desirable Minimoog sound.

- Urs

[Urs]

The demos seem way more aggressive than Diva.

Actually, hmmm, I was referencing the teaser video. Some audio demos have sounds that are distorted beyond-Minimoogish, but IMHO not desirable for a Minimoog clone. I think we'd rather work on a diode ladder instead.

[asozial0]

The demos seem way more aggressive than Diva.

Actually, hmmm, I was referencing the teaser video. Some audio demos have sounds that are distorted beyond-Minimoogish, but IMHO not desirable for a Minimoog clone. I think we'd rather work on a diode ladder instead.

safe to say that DIVA won't get that aggressive/overdriven? classic emulation vs "analogue" on steroids?

[Urs]

safe to say that DIVA won't get that aggressive/overdriven? classic emulation vs "analogue" on steroids?

To the contrary. I reckon that the Bite filter - if not driven to death - can sound fairly aggressive.

I think that overdriven isn't necessarily agressive. I think this is a common misperception of analogue sound.

That said, I'd still be happy to dissect a more aggressive sounding Minimoog.