Emulations - Architecture vs. real emulation

[kybernaut]

tomg, why do you say we need an infinite sample rate? It must just be high enough that

1) basic waveforms are reproduced with a high enough bandwdith (with regard to the human ear)

2) filters can be opened wide enough, so that they dont hearably color or dampen the waveforms anymore (most virtual analog synth are VERY weak here!)

3) crossmodulation, oscillator sync and such are fully captured and don't cause aliasing in the "wanted band"


did I forget something?

OK, we still ned accurate algorithms to reproduce the non-linearities... Of course, everything that has to do with distortion in the widest sense needs high sampling frequencies... But infinate??

[shamann]

Doesn't 24-bit depth allow you enough computational facilities to map down to the molecular level or some such?

[tomg]

that is pretty much pure nonsense. An "infinite sample rate"? Once you get past the ability of the human ear/brain to discern, it doesn't matter. what exactly that point is, I don't know, but to claim that it is infinite is ridiculous.

Plus, in fact, Vax Modular sounds fantastic. I'm not concerned with whether I or someone else could differentiate it from a real analog. It certainly sounds better than many analogs, despite its 'finite' sample rate.

You are free to believe what you want. It doesn't change the fact that software is locked at a finite sample rate and hardware runs at near the speed of light. C++ is too slow to recreate the speed of electrons. ergo... It's not even in the same ballpark. Sounds pretty good.. but not the same.

Regards

[kybernaut]

If you want to know more about emulating actual circuitry, you might want to check my paper on emulating the Moog filter: http://dafx04.na.infn.it/WebProc/Proc/P_061.pdf

Hi antti,

are you planning on a complete Mini emulation of your own? I really like your ASynth and think a fully blown antti Mini would be worth it. What do you think?


kybernaut

What exactly would it be worth?


I think, that's finally the question. But I could be wrong.

.

We have two serious attempts to do native minimoog emulation right now. Both have their strenghts and weaknesses.

It would just be interesting to see a third approach, don't you think? And antti's work sure sounds promising.

Also, I think this whole circuit modeling thing is a hot topic at the moment....

[kybernaut]

...hardware runs at near the speed of light.

btw. this is only theoretically true. All analog gear has a limited bandwidth, as well. A good analog mixing console has some 150 kHz and that's it...

[tomg]

...hardware runs at near the speed of light.

btw. this is only theoretically true. All analog gear has a limited bandwidth, as well. A good analog mixing console has some 150 kHz and that's it...

My point is that at the most basic level (the saw/square wave) software compromises. Everything after that is a cover-up.

Regards

[kybernaut]

...hardware runs at near the speed of light.

btw. this is only theoretically true. All analog gear has a limited bandwidth, as well. A good analog mixing console has some 150 kHz and that's it...

My point is that at the most basic level (the saw/square wave) software compromises. Everything after that is a cover-up.

Regards

I see... but analog oscillators are also practically limited. Just look at the edges of a square wave on a scope... they have a finite rise time.

BTW. Did you ever hear Creamware's synths? Their Pro-One and Minimax etc. have a perfect spectral reproduction of the basic waveforms.

[tomg]

tomg, why do you say we need an infinite sample rate? It must just be high enough that

I just did this in another thread...

That's not exactly true.. If you look at hardware on a scope you can see that it's on/off state occurs in exactly the same moment. (near speed of light) If you look at any softsynth there is a lag (delay) that causes all saw/square waves to lean in/out (like they want to be triangular) This slight lag (due to sample rate) removes the sharp edge and buzz-saw sound created by hardware oscillators.

This is the reason distortion has become so popular. It adds little spikes to the wave to sharpen them up. It's not the same and won't generate the same harmonics. (won't sound the same)

The same is true of VCF oscillation the after-ring (or quality) of the filter sound is also compromised.

Software does the best it can and that's pretty good but it's not the same and may never be.

Regards

[kybernaut]

I hear you -- and theoretically you are right. But I'm still sure that indistinguishable(!) analog emulation is possible and probably will be common in a few years.

regards...

[tomg]

I see... but analog oscillators are also practically limited. Just look at the edges of a square wave on a scope... they have a finite rise time.

BTW. Did you ever hear Creamware's synths? Their Pro-One and Minimax etc. have a perfect spectral reproduction of the basic waveforms

Software just adds more to the lag. (can't handle the speed of a hardware VCO)

No I haven't looked at Creamware. They may have the magic bullet but I'm skeptical at best.

I use software all the time it's a lot more that just a toy. In some way's it's better than the hardware it emulates but does it sound exactly the same....No So I use haardware all the time too. Easy

Regards

[kybernaut]

No I haven't looked at Creamware. They may have the magic bullet but I'm skeptical at best.

The just run their algorithms at a huge internal frequency...

[ross g]

just to throw in my two bits... i demoed the korg legacy ms20 and thought it was very close...i never owned one but a friend had one for about 20 years and i played it a ton... from what i remember korg claimed to model the actual transistors and such.. i think it, without doing an A/B comparison that it is worth purchasing, i actually like the sound of the new digital synths from what i've heard, and the price for analog seems to be dropping a bit because of all the powerful digi stuff... i'm covered with real analog, i guess that's the best approach imo, use the stuff that's there....

rg

[tomg]

I hear you -- and theoretically you are right. But I'm still sure that indistinguishable(!) analog emulation is possible and probably will be common in a few years.

regards...

Keyboard just did another shootout. Everybody could still recognize the hardware but the more complex the patch the harder it was to tell. In a mix I figure next to impossible

In some ways we are already there but at the most basic level where it counts the most there is a still a big difference.

Regards

[droolmaster0]

that is pretty much pure nonsense. An "infinite sample rate"? Once you get past the ability of the human ear/brain to discern, it doesn't matter. what exactly that point is, I don't know, but to claim that it is infinite is ridiculous.

Plus, in fact, Vax Modular sounds fantastic. I'm not concerned with whether I or someone else could differentiate it from a real analog. It certainly sounds better than many analogs, despite its 'finite' sample rate.

You are free to believe what you want. It doesn't change the fact that software is locked at a finite sample rate and hardware runs at near the speed of light. C++ is too slow to recreate the speed of electrons. ergo... It's not even in the same ballpark. Sounds pretty good.. but not the same.

Regards

You are free to be a moron also. Logically, the issue is obvious, and that is what is the ability of the human being to discern minute differences in quality. Beyond that it makes no difference. The speed of electrons (which is not infinite) is irrelevant. You don't know what you're talking about. The only way this speed of hardware could conceivably make a difference is if human beings could discern every possible level of detail lost when data is sampled. That's the case you need to make.

[tomg]

that is pretty much pure nonsense. An "infinite sample rate"? Once you get past the ability of the human ear/brain to discern, it doesn't matter. what exactly that point is, I don't know, but to claim that it is infinite is ridiculous.

Plus, in fact, Vax Modular sounds fantastic. I'm not concerned with whether I or someone else could differentiate it from a real analog. It certainly sounds better than many analogs, despite its 'finite' sample rate.

You are free to believe what you want. It doesn't change the fact that software is locked at a finite sample rate and hardware runs at near the speed of light. C++ is too slow to recreate the speed of electrons. ergo... It's not even in the same ballpark. Sounds pretty good.. but not the same.

Regards

You are free to be a moron also. Logically, the issue is obvious, and that is what is the ability of the human being to discern minute differences in quality. Beyond that it makes no difference. The speed of electrons (which is not infinite) is irrelevant. You don't know what you're talking about. The only way this speed of hardware could conceivably make a difference is if human beings could discern every possible level of detail lost when data is sampled. That's the case you need to make.

Regards