Was the Roland D50 the first VA Modelling Synth?

[jupiter8]

...It all comes down to how strict you are with the terms.

With genuine respect for the points you make and to the other informative posts, I don't agree. There is absolutely no ambiguity here. The word 'Virtual' in Virtual Analogue literally means that the analogue circuitry is actually modelled in the synth's computer and the D50 does not do this. Virtual Analogue synths actually behave as an analogue synthesizer does (or attempts to!).

There's absolutely 100 % no way that the Nord Lead etc. used component modelling in any modern sense of the word. Absolutely no way. Maybe the envelopes (cause that's pretty easy to do) but the oscillators and filters are just bog standard algorithms that has been around for a very long time.

Though it is an interesting question in itself,what is component modelling ?
A VCA in it's purest form is simply a multiply. Is that component modelling ? I think not. So at what level is it component modelling ? The whole component modelling is mostly marketing nonsense if you ask me.

The D50 behaves just like an analog subtractive synth (and then some but that is besides the point) but it isn't analog,so what should you call it ?

[russell.mcclellan]

Nice points by all. It's a very fine line, and since a lot of this is proprietary, secretive stuff, it would be hard to precisely classify synths even if the definition of "Virtual Analog" were clear. You also have to remember how much of a role marketing plays in the definition of terms.

Like others, I'm actually very interested in how the D-50 TVF worked. If anyone has any clues or leads, I'd love to hear them . For those of you who haven't heard the D-50 TVF with resonance, it's certainly quite a different sound than one would expect from either a simplistic digital filter (as on the TG-77) or an analog model. I did check a couple of Roland patents from that period, but none seemed to describe the filter algorithms. Perhaps it really was simply PD or FM with a different name to avoid an infringement suit.

[Automatic Gainsay]

This is far less complicated than everyone is making it to be.

First of all, I don't remember the term "virtual analog" being applied to ANY synthesizer before it was applied to software synths which were physically modeling analog synthesizers. As such, I agree with all of the people who have made that connection.

Hardware-wise, I agree with whomever pointed to the Prophecy. While I won't call it "virtual analog," because it isn't virtual, really... it was the first hardware I remember to physically model various instrument sounds.

And to whomever said that the Roland JD was a ROMpler simply because it had sampled waveforms... gahhhh. ROMplers aren't about sampled waveforms, they're about sampled sounds for the purpose of imitation. Synthesizers aren't ROMplers just because they have sampled waveforms. [god please don't let that debate start up again]

MOST OF ALL, the D-50 was NEVER intended to be a synth to simply imitate other instruments... it was a new synthesizer meant to do new synthesizer stuff. Sure, it had some actual instrument sounds on it, and some synthesizery sounds. But it is not analog, it doesn't sound like analog, and it wasn't meant to (as in that was its intent, and its marketing aim) imitate analog in any way, despite being able to make analogish sounds (like any other damned synth).
"Subtractive" is a misleading and stupid term, even though Moog himself used it. Defining a synth by what the filter does is particularly retarded because there is SO much more going on in a synth than just what the filter is doing. MOST analog synths are also additive, if they have multiple oscillators. In any case "subtractive" is not equal to "analog." Digital synths work on subtractive principles, too. "Analog" does not mean "has a low pass filter."

This whole argument is basically revisionist history.

[nvbrkr]

So to be clear on this - a virtual analogue synthesizer actually attempts to behave as an analogue synthesizer through a mathematical model

Correct.

The word 'Virtual' in Virtual Analogue literally means that the analogue circuitry is actually modelled in the synth's computer.

Incorrect.

The word "virtual" does not imply such a thing by definition. A piece of software or hardware being called "virtual" means It aims to reproduce the functional properties of a certain environment and not its properties in the sense that they actually exist in the physical world. It's all about the behaviour (though not in any "psychological" sense, unless we're including the user-level experience of it in the term itself by default). Any attempts at convincingly recapturing the characteristics of a specifc, sought-after piece of equipment are secondary. For that matter, "modelling" doesn't mean there exists some sort of a replica of a circuitry in the code either. It refers to the various processes the design team has done in order to study the item whose characteristics they are aiming to reproduce and how they attempted to reproduce it. We all know how rudimentary the "modelling" e.g. on amp simulators like the Line6 POD really is.

So you need to be clear - Virtual Analogue synthesizers only 'happen to use digital techniques' to model analogue ones - that fact that they use digital techniques is almost by the by - perhaps in the future we'll use quantum computers or biological computers to implement the model - but if the model is the same then it'll be the same and sound essentially the same as out current digital implementations.

What do biological computers have to do with this for that matter? Are you sure you're not throwing around terms you don't necessarily understand?

[knolan]

Its not your call.

why not?

such imagined authority

please explain the full d 50 synth engine,
including the generation of the saw/squ waveforms and the implementation of the filter.

The point I'm making that I believe is being missed is that a virtual analogue synthesizer models circuits, not oscillator wave types. It may be a mathematically calculated sawtooth - but that’s a digital oscillator. But, for the likes of the Arturia Minimoog (and indeed all Virtual Analogue synthesizers (again its all in the name)), actual circuits are modelled - and those virtual circuits generate oscillators. In other words, its transistors, resistors, diodes, capacitors and so on that are modelled - which happen to generate the various analogue features of an actual analogue synthesizer, but in a virtual world. This is a totally different thing to what's going on in the D50, however you describe it.

The 'not your call' statement is not a point of authority. This charge is very common these days, as if it has to be that all facts are wide open to debate and that all opinions are valid. Not all facts are open to debate. And - not all opinions are equally valid. And that's the case here. You mighn't like it but its the way it is. You cannot arbitrarily call a well understood synthesis method such as virtual analogue synthesis anything you like. It’s just not on; and as said repeatedly - too many people are calling any synthesizer they want a virtual analogue when in reality it is not. This has nothing to do with authority, it has nothing to do with me or you, but you must accept that if someone is going to start calling a Casio CZ synthesizer a virtual analogue synthesizer they are factually incorrect.

I believe there is confusion here between analogue synthesis and subtractive synthesis; and as was pointed out by another poster, between physical modelling and virtual analogue. These are all distinctive terms in their own right. In fact I'd go as far as saying that it is NOT clear that the AN1x is a virtual analogue synthesizer. I believe it is more accurately a virtual synthesier that models the waves and filters and so on of a subtractive synthesizer. But it's not clear if Yamaha use an analogue circuit model in this synthesizer (perhaps they did but I haven't seen reference to that).

In fact even in the latest Music Tech Focus series on synthesis they called Trillian a Virtual Instrument when it is not. The language is just getting too sloppy and is betraying the synthesis mechanism underneath. Such sloppiness has implications - confusion by those using them, confusion to all new comers (and rest assured there are more people coming to synthesis today than ever before); dishonesty in second hand sales; confusion in resource planning and all round unprofessional. And - such loose terminology betrays the huge efforts that the likes of Clavia, Korg and Arturia have put in to actually model analogue circuits. I've seen plenty of venom on this forum against Arturia on what I feel can often be a misunderstoond basis. The likes of Arturia synthesizers (and the Korg Legacy range) are strewn with character that other digital synths like the JD800 cannot match because they model actual analogue electronics. They are a marvel of modern synthesis and a class all of their own.

Regarding the poster just above this one "nvbrkr" - my point on other types of computers is this - the character and sound of, for example, the Korg Legacy PolySix (which sounds very like the original by the way) is based on the actual model Korg have come up with to emulate the Polysix analogue electronics. So, in the future if we use other types of computers such as biological computers but with he same virtual analogue Polysix model - then it will have the same sound and character. In other words - the sound of a virtual analogue synthesizer is (mostly) down to how it models the analogue circuitry and is mostly independent of the computer system running that model (apart form ADCs and so on). I pointed this out because it appeared to me that some people though that, for example, a digitally generated sawtooth wave is a virtual analogue sawtooth wave (which it isn't). But I know you already know that and that you are splitting hairs on this to be argumentative !!!


Kevin.

[adamstan]

In other words, its transistors, resistors, diodes, capacitors and so on that are modelled

No, they aren't. Somebody pointed that out already. Just run a PSPICE simulation of simple VCO - it takes too much time to be done in realtime even on current fast PCs. When manufacturers say that their digital oscillators or filters are "carefully modelled after analogue ones" they just mean, that they've used standard osc/filter algorithms and tweaked their parameters to get similar results. There is no actual circuit simulation in VA synths.

[StepLogik]

In other words, its transistors, resistors, diodes, capacitors and so on that are modelled

No, they aren't. Somebody pointed that out already. Just run a PSPICE simulation of simple VCO - it takes too much time to be done in realtime even on current fast PCs. When manufacturers say that their digital oscillators or filters are "carefully modelled after analogue ones" they just mean, that they've used standard osc/filter algorithms and tweaked their parameters to get similar results. There is no actual circuit simulation in VA synths.

Thank you for making a point I was about to

I've been waiting on true analog circuit emulation for a very long time and I still think we've got quite a ways to go before there is enough CPU horsepower to do this on typical consumer computers.

Even with the dedicated specialty processors, this level of modeling is not really feasible. I'm annoyed when I see marketing literature that implies they are modeling down to the analog component level. Simply not true.

[mute]

In other words, its transistors, resistors, diodes, capacitors and so on that are modelled

No, they aren't. Somebody pointed that out already. Just run a PSPICE simulation of simple VCO - it takes too much time to be done in realtime even on current fast PCs. When manufacturers say that their digital oscillators or filters are "carefully modelled after analogue ones" they just mean, that they've used standard osc/filter algorithms and tweaked their parameters to get similar results. There is no actual circuit simulation in VA synths.

Thank you for making a point I was about to

I've been waiting on true analog circuit emulation for a very long time and I still think we've got quite a ways to go before there is enough CPU horsepower to do this on typical consumer computers.

Even with the dedicated specialty processors, this level of modeling is not really feasible. I'm annoyed when I see marketing literature that implies they are modeling down to the analog component level. Simply not true.

Not feasible? o_O no no, you exaggerate heavily. A shitpile of work not many people are willing to dedicate themselves to for marginal difference? Yes. This is what component modelling is and it gets better all the time, although yes.. many ppl abuse the term generically.

There are products out there that do precisely that, the most recent/popular/best example is FXPansion's DCAM (Discrete Component Analogue Modelling). That 's DCAM is ALL about, nothing more, nothing less, and they've presented it as growing library of Models so they can release multiple synths based on their DCAM platform. http://www.fxpansion.com/index.php?page=62 ... its now over a year old and when it was released it made quite a bit of news, so im a bit suprised you dont already know of it.

[I12]

Its not your call.

why not?

such imagined authority

please explain the full d 50 synth engine,
including the generation of the saw/squ waveforms and the implementation of the filter.

The point I'm making that I believe is being missed is that a virtual analogue synthesizer models circuits, not oscillator wave types. It may be a mathematically calculated sawtooth - but that’s a digital oscillator. But, for the likes of the Arturia Minimoog (and indeed all Virtual Analogue synthesizers (again its all in the name)), actual circuits are modelled - and those virtual circuits generate oscillators. In other words, its transistors, resistors, diodes, capacitors and so on that are modelled - which happen to generate the various analogue features of an actual analogue synthesizer, but in a virtual world. This is a totally different thing to what's going on in the D50, however you describe it.

The 'not your call' statement is not a point of authority. This charge is very common these days, as if it has to be that all facts are wide open to debate and that all opinions are valid. Not all facts are open to debate. And - not all opinions are equally valid. And that's the case here. You mighn't like it but its the way it is. You cannot arbitrarily call a well understood synthesis method such as virtual analogue synthesis anything you like. It’s just not on; and as said repeatedly - too many people are calling any synthesizer they want a virtual analogue when in reality it is not. This has nothing to do with authority, it has nothing to do with me or you, but you must accept that if someone is going to start calling a Casio CZ synthesizer a virtual analogue synthesizer they are factually incorrect.

I believe there is confusion here between analogue synthesis and subtractive synthesis; and as was pointed out by another poster, between physical modelling and virtual analogue. These are all distinctive terms in their own right. In fact I'd go as far as saying that it is NOT clear that the AN1x is a virtual analogue synthesizer. I believe it is more accurately a virtual synthesier that models the waves and filters and so on of a subtractive synthesizer. But it's not clear if Yamaha use an analogue circuit model in this synthesizer (perhaps they did but I haven't seen reference to that).

In fact even in the latest Music Tech Focus series on synthesis they called Trillian a Virtual Instrument when it is not. The language is just getting too sloppy and is betraying the synthesis mechanism underneath. Such sloppiness has implications - confusion by those using them, confusion to all new comers (and rest assured there are more people coming to synthesis today than ever before); dishonesty in second hand sales; confusion in resource planning and all round unprofessional. And - such loose terminology betrays the huge efforts that the likes of Clavia, Korg and Arturia have put in to actually model analogue circuits. I've seen plenty of venom on this forum against Arturia on what I feel can often be a misunderstoond basis. The likes of Arturia synthesizers (and the Korg Legacy range) are strewn with character that other digital synths like the JD800 cannot match because they model actual analogue electronics. They are a marvel of modern synthesis and a class all of their own.

Regarding the poster just above this one "nvbrkr" - my point on other types of computers is this - the character and sound of, for example, the Korg Legacy PolySix (which sounds very like the original by the way) is based on the actual model Korg have come up with to emulate the Polysix analogue electronics. So, in the future if we use other types of computers such as biological computers but with he same virtual analogue Polysix model - then it will have the same sound and character. In other words - the sound of a virtual analogue synthesizer is (mostly) down to how it models the analogue circuitry and is mostly independent of the computer system running that model (apart form ADCs and so on). I pointed this out because it appeared to me that some people though that, for example, a digitally generated sawtooth wave is a virtual analogue sawtooth wave (which it isn't). But I know you already know that and that you are splitting hairs on this to be argumentative !!!


Kevin.

I'm applying for a grant next year
can you ghost write it for me?

[nSCOURGE]

Regarding the poster just above this one "nvbrkr" - my point on other types of computers is this - the character and sound of, for example, the Korg Legacy PolySix (which sounds very like the original by the way) is based on the actual model Korg have come up with to emulate the Polysix analogue electronics. So, in the future if we use other types of computers such as biological computers but with he same virtual analogue Polysix model - then it will have the same sound and character.

Not necessarily.

As this says nothing about those models being a dependant variable(consider the real mathematical consequences between a fixed vs floating point approach).

I have the V-station native(floating point), and the Powercore version(fixed point double-precision), where the differences become obvious(as subtle as they may present to cursory investigation).

[nSCOURGE]

Not feasible? o_O no no, you exaggerate heavily. A shitpile of work not many people are willing to dedicate themselves to for marginal difference? Yes. This is what component modelling is and it gets better all the time, although yes.. many ppl abuse the term generically.

There are products out there that do precisely that, the most recent/popular/best example is FXPansion's DCAM (Discrete Component Analogue Modelling). That 's DCAM is ALL about, nothing more, nothing less, and they've presented it as growing library of Models so they can release multiple synths based on their DCAM platform. http://www.fxpansion.com/index.php?page=62 ... its now over a year old and when it was released it made quite a bit of news, so im a bit suprised you dont already know of it.

And what explicit synth(s) did they model these components from?

If this information is not forthcoming, I might begin to question if it is perhaps a strategic omission(in evading
scrutiny with respect to the fidelity/precision of their modelling techniques).

[nSCOURGE]

...It all comes down to how strict you are with the terms.

With genuine respect for the points you make and to the other informative posts, I don't agree. There is absolutely no ambiguity here. The word 'Virtual' in Virtual Analogue literally means that the analogue circuitry is actually modelled in the synth's computer

That ignores popular convention(which you admittedly lament) that observes a 'VA' to mean any synth which demonstrates behaviors which are characteristic of analog synthesizers.

But, it is trivial to argue that preference should be given to the *higher-level* behavior of any emulation, given it's ad-hoc implications.

So, Justifying any proposed VA's validity inevitably becomes a hair-splitting excercise in infinite regression(where modeling 'uncertainties' can only be unified/resolved through a priori assumptions).

And, the devil is in the details - split enough hairs, and you will falsify any analogy.

Interestingly, your definition should serve to disqualify alot of heretofor commonly acknowledged 'VA' synths from the late 90's(some of which *you* are advocating as genuine VA).

[nSCOURGE]

Personally, if for example I saw somebody selling a Casio CZ synth on ebay as a virtual analogue synthesizer I'd flag the con to ebay themselves.

[nvbrkr]

"nvbrkr" - my point on other types of computers is this - the character and sound of, for example, the Korg Legacy PolySix (which sounds very like the original by the way) is based on the actual model Korg have come up with to emulate the Polysix analogue electronics. So, in the future if we use other types of computers such as biological computers but with he same virtual analogue Polysix model - then it will have the same sound and character. In other words - the sound of a virtual analogue synthesizer is (mostly) down to how it models the analogue circuitry and is mostly independent of the computer system running that model (apart form ADCs and so on). I pointed this out because it appeared to me that some people though that, for example, a digitally generated sawtooth wave is a virtual analogue sawtooth wave (which it isn't). But I know you already know that and that you are splitting hairs on this to be argumentative !!!

Okay, sure. You're still throwing around vague references to forms of techonology that very few people understand. I don't think that is directly related to the subject matter at all. If you think it's relevant then by all means continue doing it.

That 's DCAM is ALL about, nothing more, nothing less, and they've presented it as growing library of Models so they can release multiple synths based on their DCAM platform. http://www.fxpansion.com/index.php?page=62 ...

Google brought up this:
http://www.fxpansion.com/forum/viewtopi ... 87da6326e1

"DCAM is a set of approaches, technologies, knowledge and experience, rather than one hard-and-fast platform. To grossly oversimplify things - we use QUCS (basically the same thing as SPICE) to analyse the circuits, then build mathematical approximations to those that will work in a realtime, discrete context. We of course refer to the actual circuit itself, as well as the schematic and the model, to make sure everything tallies. There is a certain amount of judgement involved as to which parts of the circuit, or which of its behaviours, are most important to the sound, as even with modern CPUs it's not possible to make a perfect model work in real-time at 44K. Yet."

From what I can pick up from that, that doesn't mean they have simulated all the components themselves. It still seems to rely on approximations of the overall output. It's not terribly clear either how the "most important parts" of the circuit should function correctly in the "virtual model" if you're not doing the rest as well. However, if these guys aren't doing that now - how likely is it that Arturia were doing anything of the kind with their old Minimoog / CS-80 emulations?

[edit] I checked out the sound samples of the DCAM synths and I guess they sounded a little bit different from most softsynths. They do not sound like analog to me, but I guess there is still more motion to the sound than usually.

[OriginalJambo]

Yes, I'd say so.

Despite Roland's original intentions I'd say that since the D-50 technically was the first digital "subtractive" synthesiser that generated its output in real time, it should therefore be considered close enough in design to modern VAs to qualify.