Arturia V Collection 6

[wagtunes]

What knowledge? You have no knowledge at all

Oh for goodness sake, enough already. You don't agree, that's fine. Neither of you are going to change their views, so continuous posts on this are just creating more pointless noise that's distracting us from important moaning about upgrade pricing...

LMAO

Damn, I want this thing SOOOOOO bad.

[ghettosynth]

But there is one fundamental problem: It is not possible to mathematically prove the accuracy of these emulations.

That isn't the claim, you should work on your reading comprehension.

Asking for an answer was just rhetorical because of your steep claim.

It wasn't a steep claim, you just don't get it.

I thought that you were done?

[ghettosynth]

What knowledge? You have no knowledge at all

Oh for goodness sake, enough already. You don't agree, that's fine. Neither of you are going to change their views, so continuous posts on this are just creating more pointless noise that's distracting us from important moaning about upgrade pricing...

LMAO

Damn, I want this thing SOOOOOO bad.

This is what Arturia is counting on. They want your desire to overwhelm your sense of valuation.

[chk071]

To get back to the point: HOW can you calculate the accuracy of an emulation mathematically? I'd really like to know, because i wonder how you want to accomplish that. If you mean that you can judge how accurate the circuit modelling is to the real deal, then, yeah, there's surely a point to that, but, mathematically measuring something you have no idea about, code wise? How's that going to work?

Maybe i misunderstood you there, and that's not how you meant it. Not sure.

[ENV1]

As i mentioned in my post above the RAM use with muliple instances is much less here if the GUis are closed (with 10 instances an average between around 80 to 100 MB).

This also depends on the host and whether the instances are being run standalone or not.

Here are some VSTHost data:

- VSTHost alone: 13MB

- With 1 instance, UI closed: 74MB

- With 1 instance, UI opened: 258MB, no change when UI is closed again

- With 2 instances, UIs closed: 127MB

- With 2 instances, UIs opened: 392MB, 368MB when UIs are closed again

- With 3 instances, UIs closed: 181MB

- With 3 instances, UIs opened: 522MB, 498MB when UIs are closed again

Didnt feel like doing more because we get the picture.


Note: If more then 1 instance is loaded into VSTHost, VSTHost locks up on exit and has to be force-killed.

(Checked 5 times and it happened every time.)

[wagtunes]

To get back to the point: HOW can you calculate the accuracy of an emulation mathematically? I'd really like to know, because i wonder how you want to accomplish that. If you mean that you can judge how accurate the circuit modelling is to the real deal, then, yeah, there's surely a point to that, but, mathematically measuring something you have no idea about, code wise? How's that going to work?

Maybe i misunderstood you there, and that's not how you meant it. Not sure.

I'm not a techie (not even close) but I'm gonna take a stab at this.

The guys who make these software emulations take the synth in question and tear it apart to see what's inside. They then run all these "tests" to see what the synth is "physically" doing when you press a key. I have no idea what kinds of tests are involved but I'm assuming there are things you can do to evaluate the operation of a synth.

Then, after they've compiled all this data on oscillators, filters and whatever, they then take their computer code, in whatever language they're programming in, and try to replicate this physical activity inside the code. And then I guess there are some instruments they can run the results through to see how close they came to the original.

Then I assume, because of the limitations of PC processing power, they have to make some decisions on what to keep and what to eliminate in order to make it so that a typical user can actually run the software. How many compromises are made and the degree of each compromise is what ultimately determines how close they actually come to the original.

This is just a logical walk through and I'm sure nothing that you haven't actually figured out for yourself.

Oh, and of course it's always possible that they also access schematics of filter models, etc., to also recreate these simulations.

I don't think anybody here can really answer this question in great depth unless they've actually gone through the process of creating a synth emulation. I'm sure somebody like Urs can probably explain the process and how to measure the accuracy. I'm also sure that, for the average person, it would make their eyes glaze over.

I know it would mine.

[martinjuenke]

What knowledge? You have no knowledge at all

Oh for goodness sake, enough already. You don't agree, that's fine. Neither of you are going to change their views, so continuous posts on this are just creating more pointless noise that's distracting us from important moaning about upgrade pricing...

You‘re sooooo right!
Stop your alpha boys struggle!

[chk071]

To get back to the point: HOW can you calculate the accuracy of an emulation mathematically? I'd really like to know, because i wonder how you want to accomplish that. If you mean that you can judge how accurate the circuit modelling is to the real deal, then, yeah, there's surely a point to that, but, mathematically measuring something you have no idea about, code wise? How's that going to work?

Maybe i misunderstood you there, and that's not how you meant it. Not sure.

I'm not a techie (not even close) but I'm gonna take a stab at this.

The guys who make these software emulations take the synth in question and tear it apart to see what's inside. They then run all these "tests" to see what the synth is "physically" doing when you press a key. I have no idea what kinds of tests are involved but I'm assuming there are things you can do to evaluate the operation of a synth.

Then, after they've compiled all this data on oscillators, filters and whatever, they then take their computer code, in whatever language they're programming in, and try to replicate this physical activity inside the code. And then I guess there are some instruments they can run the results through to see how close they came to the original.

That's strict circuit modelling. There are other ways to do oscillators, filters, and what not in DSP though. E.g. something like Waldorf Nave doesn't do any kind of circuit modelling, it's all "classic" DSP, thus takes only a fraction of what these monster emulations take, CPU wise. After all, they had to run it on an iPad.

[murnau]

"The accuracy" is more then the sum of all parts = Mathematics ends where art is setting in.

[wagtunes]

To get back to the point: HOW can you calculate the accuracy of an emulation mathematically? I'd really like to know, because i wonder how you want to accomplish that. If you mean that you can judge how accurate the circuit modelling is to the real deal, then, yeah, there's surely a point to that, but, mathematically measuring something you have no idea about, code wise? How's that going to work?

Maybe i misunderstood you there, and that's not how you meant it. Not sure.

I'm not a techie (not even close) but I'm gonna take a stab at this.

The guys who make these software emulations take the synth in question and tear it apart to see what's inside. They then run all these "tests" to see what the synth is "physically" doing when you press a key. I have no idea what kinds of tests are involved but I'm assuming there are things you can do to evaluate the operation of a synth.

Then, after they've compiled all this data on oscillators, filters and whatever, they then take their computer code, in whatever language they're programming in, and try to replicate this physical activity inside the code. And then I guess there are some instruments they can run the results through to see how close they came to the original.

That's strict circuit modelling. There are other ways to do oscillators, filters, and what not in DSP though. E.g. something like Waldorf Nave doesn't do any kind of circuit modelling, it's all "classic" DSP, thus takes only a fraction of what these monster emulations take, CPU wise. After all, they had to run it on an iPad.

Well, regardless of what the process is, there has to be a way for them to look at what they made and look at what they emulated it from and determine how close they came other than with just their ears. There has to be some kind of tool or tools that can measure this accuracy in order for us to declare that the synth(s) is accurate.

Otherwise, all of this is pointless to discuss.

Again, I'm sure somebody like Urs can enlighten us on this subject.

[chk071]

To get back to the point: HOW can you calculate the accuracy of an emulation mathematically? I'd really like to know, because i wonder how you want to accomplish that. If you mean that you can judge how accurate the circuit modelling is to the real deal, then, yeah, there's surely a point to that, but, mathematically measuring something you have no idea about, code wise? How's that going to work?

Maybe i misunderstood you there, and that's not how you meant it. Not sure.

I'm not a techie (not even close) but I'm gonna take a stab at this.

The guys who make these software emulations take the synth in question and tear it apart to see what's inside. They then run all these "tests" to see what the synth is "physically" doing when you press a key. I have no idea what kinds of tests are involved but I'm assuming there are things you can do to evaluate the operation of a synth.

Then, after they've compiled all this data on oscillators, filters and whatever, they then take their computer code, in whatever language they're programming in, and try to replicate this physical activity inside the code. And then I guess there are some instruments they can run the results through to see how close they came to the original.

That's strict circuit modelling. There are other ways to do oscillators, filters, and what not in DSP though. E.g. something like Waldorf Nave doesn't do any kind of circuit modelling, it's all "classic" DSP, thus takes only a fraction of what these monster emulations take, CPU wise. After all, they had to run it on an iPad.

Well, regardless of what the process is, there has to be a way for them to look at what they made and look at what they emulated it from and determine how close they came other than with just their ears. There has to be some kind of tool or tools that can measure this accuracy in order for us to declare that the synth(s) is accurate.

I'd think that scopes and spectrometers should be the best (and only) way to determine that.

[martinjuenke]

May we call this a derailed thread?

[wagtunes]

May we call this a derailed thread?

But then wouldn't we have to do that with 90% of the threads posted here? LOL

[murnau]

Derailed? Then bring it back on track!

[jme-audio]

I'd think that scopes and spectrometers should be the best (and only) way to determine that.

Yes, and you can use sampling software, so that you can exactly measure envelope times.