Diva Vs. Real Analog

[goldenanalog]

A "reconstructed" waveform doesn't have to "look", it has to sound like the original waveform. DAC reconstruction filters have reached the point of providing far above satisfactory results for perceived (read: by humans) signals ages ago. Same with ADC antialiasing filters. All this is besides the point anyway. Everyone here runs both analog and digital synthesizers through reconstruction filters. Do you really feel like converting this thread to yet another analog vs digital medium "debate" thread? Not that I care, because this thread is a piece of shit already anyway.

You may find this disingenuous, Shy; but I agree in large part with what you are saying. But: within the context of comparing musical instruments, I think that the finer points matter. A lot.

IMO: We are coming down to the wire, Shy. DIVA could almost universally be called a superb software rendition of a analog synthesizer. So: Where do we go from here?

I've got about 100 licenses for different softs; most of the major highly-rated soft synths live across several different machines...yes, I got analog synths, digital synths, recorders and recoding software, etc. etc. like you, and most others here.

So the question here is this. Can Diva compete with actual Analog synths? I doubt that VAs will completely replace real Analog within the nexc 10 years or so, but I think Diva is probably up there with some of the good Analog synths on the market right now. Does anyone own an anlog synth like the Evolver, Little/Slim Phatty, Voyager or the like? Have you compared Diva to one of these? My guess is that Diva would hold its own against some of these synths. I don't mean to say that it's necessarily a perfect reproduction of analog sound, but it has its OWN sound that is extremely flexible and as sweet to the ears as some analog synths.

VA's just might. The only reason that I can think of why VA's won't replace analog synthesis systems may have more to do with the amount of required information, which is inherently limited in a ADA system.

[ho66it]

Have you enabled "Display realtime (RT) CPU on graph" in the Reaper performance meter? I think this means to show the maximum usage for any single core in use by Reaper. At least, this seemed to correspond to the core utilized by Diva for me. Otherwise, you can try Studio One, since this has per-core CPU monitor.

Thank you for the advice. I know that I did not enable the realtime display in Reaper. I'll try it. However, now when I right click on the Performance Meter to call the context menu so that I can enable it, the context menu does not appear. I'll re-install Reaper and see if that changes...
I'll take a look at Studio One as well. I'm sure I can certainly live without seeing the realtime cpu hit though. After all, Diva still sounds great!

[Old Crab]

You realize that you can have a different setting for offline rendering right? That is, DIVA can be as divine as you like once you decide to record her:)

True enough, thanks.

[Aroused by JarJar]

Back OT (DIVA vs. Analog):

Quantity of information:

44,100/48,000/96,000 points vs: INF (continuous; no reconstruction filter used or required-no breakdown of the wave-shape at the boundaries)

1. Information is physical stuff, so the quantity of information in any analog signal over a fixed time can not be infinite (that would break the first law of thermodynamics).

2. After DA conversion the resultant signal is analog anyway, so has the exact same quality of continuity.

3. Not sure what any of this has to do with the relative merits of either as musical instruments.

Thanks for taking the time to respond, hakey!

3.) This is really just a coarse comparison between 2 different methods of synthesis, and the possible reasons why they inherently sound different.

1.) No matter how small of a window of time that you use to measure an analog signal, there will be a rate of change associated with that period of time-even when your time measurement window is infinitely small (in theory, at least-I recognize that there are limits in a physical system)

Immediately after conversion back to analog, the reconstructed waveform looks like a series of discrete values due to sample/hold circuitry 'freezing' the last given value. Rate of change *inside* a sample period is fixed: zero. *Between* sample periods: asymptotic.

2.) Reconstruction filters are imperfect; therefore, there is distortion introduced when a wave is 'reconstructed'. This is readily apparent at the boundary-if you were to attempt to reproduce an 10khz analog wave (any shape)using 44.1khz as the sampling frequency, the reconstructed waveform would probably not look like the original.

An analog synthesizer recorded to digital on anything but the most crap equipment still sounds like an analog synthesizer. Obviously higher sampling and bit rates, and better ADDA filtering are going to capture more of the signal, but it is clearly not sampling limitations which are the cause of things sounding "not analog". If sampling limitations were at fault, it would not be possible to capture analog sound in recordings. Sure an analog synthesizer sounds better in real life, what doesn't? But the great bulk of distinguishing characteristics survives even 16/44.1 recording/playback and digital processing.

[goldenanalog]

Thanks, Aroused by jarjar for your thoughts! If what you are saying is true, then the natural question is: Why doesn't digital sampling of notes across a given analog synth adding round-robin info; then playback of those samples solve the problem? And what about live situations for the performer? He or she knows when they are playing the real deal. Let's see where we are in a few years after other Dev's have started catching up to Urs, and Urs has pushed us further ahead.

[deastman]

Thanks, Aroused by jarjar for your thoughts! If what you are saying is true, then the natural question is: Why doesn't digital sampling of notes across a given analog synth adding round-robin info; then playback of those samples solve the problem? And what about live situations for the performer? He or she knows when they are playing the real deal. Let's see where we are in a few years after other Dev's have started catching up to Urs, and Urs has pushed us further ahead.

Sure, the samples may sound a lot like the real thing. But that is a static snapshot of a synth, not an actual synth full of variable parameters. If you you want to modify the filter cutoff after the fact, you're going to be stuck with the digital filter rather than the analog one. Also keep in mind that certain aspects of the real analog synth, such as oscillator phase relationships, may not always translate properly to playback of a static set of samples.

[Aroused by JarJar]

Well the most obvious thing is that the signal from an analog machine is always subtley changing. So you'd need long as well as accurate samples. And an analog synth usually is not played like an organ- so even if you had perfectly accurate oscillator emulations, everything else would have to modelled.

But keep in mind that when people say "analog", they usually actually have an image in their mind that is not in fact "equal" to "analog". For example, sometimes when people hear MOTM or Cwejman, which are analog synths at the very top, they complain that they sound "digital". Clearly these people have an image of analog that is noisy and dark. And some people, when they get a chance to play a real noisy dark thick analog are disappointed- it turns out that their conception of "analog" is actually glaring and squealing 1990's VA sounds.

[hakey]

Quantity of information:

44,100/48,000/96,000 points vs: INF (continuous; no reconstruction filter used or required-no breakdown of the wave-shape at the boundaries)

1. Information is physical stuff, so the quantity of information in any analog signal over a fixed time can not be infinite

No matter how small of a window of time that you use to measure an analog signal, there will be a rate of change associated with that period of time-even when your time measurement window is infinitely small

You're original claim - the one I responded to - was that an analog signal of finite length can carry an infinite quantity of information.

You are now talking about infinite resolution; that is not a proof of your infinite quantity of information claim, and is not anyway possible (because of signal noise, max slew rate, the Planck length etc).

And if you want to talk about the rate of the transfer of information, surely the human auditory system (bandwidth ~50Hz to ~16kHz) is the main limiting factor?

[pdxindy]

Well the most obvious thing is that the signal from an analog machine is always subtley changing. So you'd need long as well as accurate samples. And an analog synth usually is not played like an organ- so even if you had perfectly accurate oscillator emulations, everything else would have to modelled.

But keep in mind that when people say "analog", they usually actually have an image in their mind that is not in fact "equal" to "analog". For example, sometimes when people hear MOTM or Cwejman, which are analog synths at the very top, they complain that they sound "digital". Clearly these people have an image of analog that is noisy and dark. And some people, when they get a chance to play a real noisy dark thick analog are disappointed- it turns out that their conception of "analog" is actually glaring and squealing 1990's VA sounds.

So true... analog can sound cold and 'digital'

[hakey]

Well the most obvious thing is that the signal from an analog machine is always subtley changing.

Ah yes, heat shimmer.

[Shy]

IMO: We are coming down to the wire, Shy. DIVA could almost universally be called a superb software rendition of a analog synthesizer. So: Where do we go from here?

I disagree. Even the ancient Nord Lead sounds better than it.

[hakey]

Even the ancient Nord Lead sounds better than it.

All that lovely aliasing...

[Shy]

Yes, even despite that shitty aliasing it sounds better, and NL2X minimizes it so much that it's not an issue. If Diva can't even sound as good as a 16 years old virtual analog synth, I wouldn't call it "revolutionary". It has more filter types, so that's one aspect where it's more flexible.

At least The Drop is the first software filter that really shows "revolutionary" progress in sound quality. When a synth with proper oscillators gets attached to it, that would be a revolutionary virtual analog synth.

[penguinfromdeep]

We want some audio examples to back up your claims Shy!

[Shy]

I don't need to back up anything, you can agree with my opinion or not. You can listen to music by countless well known electronic music artists to hear that synth if you want. Too bad I don't have any dry recordings of the NL2X I had except this (compared to this). Anyway, whatever.