What's the remaining gap for impeccable analog modelling today?

[Jafo]

We've been at the point where only what one had for lunch will find a difference.
Besides, any physicist well tell you that reality is digital anyway.

Please quote me fully: "We've been at the point where only confirmation bias, comb filtering, what one had for lunch and maybe phase of the Moon will find a difference."

Starsky Carr's Minimoog shootout should tell you everything you need to know about contemporary analog modelling in VSTs:

https://www.youtube.com/watch?v=S1h07ws--CM

https://www.youtube.com/watch?v=S1h07ws--CM
I'll just leave this here.[/quote]

Alas, it's not valid evidence, since it's not double-blind -- as well as being an example of confirmation bias. Yes, there are clear differences between each emulation, but there are also clear differences between each hardware Minimoog. Which emulation is the true Minimoog? Which Minimoog is the real Minimoog? (It's like asking what's the best PAF clone in a pickup forum.)

Still, it's a great comparison, and has me rethinking my position a bit; thank you. Gives me hope that I might be wrong!

[aciddose]

Well enough that there is no reliable distinction which cannot be adequately explained by confirmation bias or random factors related to the recording or listening environment.

That's simply untrue though. There has never been any reason that perfect emulations aren't possible: they have however remained almost completely impractical.

There is no possible way to make an "ideal" (as perfect as you describe) software emulation of an electronic circuit more cost-efficient than the actual hardware, even when you factor in the cost of the hardware and the fact the software can be endlessly duplicated nearly for free.

The problem is that there simply isn't enough demand in the world for the type of system and software required to provide that sort of highly accurate emulation for the true cost of the system (small scale super-computers?) and the R&D, development time & skill required.

As evidenced by your opinion that existing software is "good enough", there is truly no significant demand for highly accurate emulations to begin with! So how could the author of such an emulation possibly expect to fund their development effort with only one or two customers world-wide?

People who want such a highly accurate emulation already own the hardware, so the only question is of cost and practicality. Sure, you can create 15 instances of an emulation while you can't so easily duplicate hardware, but can you run those 15 instances all together on your inexpensive (non-super-computer) DAW in real-time? Not a chance!

It simply isn't possible to create a "more efficient" emulation due to the laws of physics!

[aciddose]

Alas, it's not valid evidence, since it's not double-blind

Evidence of what? Objective facts? Or subjective opinions? Nobody cares about subjective opinions in this discussion, otherwise we'd be arguing over how flavors of salad dressing or the wallpaper in your studio affect your mix.

What we're talking about is whether there are measurable differences between software emulations and the original hardware they claim to emulate.

Yes, there are clear differences between each emulation, but there are also clear differences between each hardware Minimoog. Which emulation is the true Minimoog? Which Minimoog is the real Minimoog? (It's like asking what's the best PAF clone in a pickup forum.)

You need to be feigning ignorance (disingenuity) in order to miss out on simple facts such as talking about the difference between an original painting vs. a print. Nowhere in the original painting are there CMYK dots or aliasing from discretization of the image.

You're changing the subject and setting up a straw-man by arguing that an average magazine reader can't detect the difference between a CMKY print of the original painting vs. a print of the print!

It takes a massive amount of knowledge and experience to become aware of the real differences and how to objectively measure them. This is like in my previous post where I'm pointing out there is little demand for such accuracy in emulations and that it is impractical to provide that accuracy even if there were sufficient demand.

I'm not talking about barely perceptible differences, I'm talking about quantifying major differences that you incorrectly attribute to different models which are in fact simply the result of a lack of any proper attempt at emulation of those properties at all.

[Jafo]

Yeesh, I wish I could keep up with you...

Alas, it's not valid evidence, since it's not double-blind

Evidence of what?

That modern emulations are good enough to reliably fool people.

Objective facts? Or subjective opinions? Nobody cares about subjective opinions in this discussion, otherwise we'd be arguing over how flavors of salad dressing or the wallpaper in your studio affect your mix.

What we're talking about is whether there are measurable differences between software emulations and the original hardware they claim to emulate.

Hmm... it occurs to me that maybe we're talking about slightly different things -- that my goals are lesser, easier to achieve.

Yes, there are clear differences between each emulation, but there are also clear differences between each hardware Minimoog. Which emulation is the true Minimoog? Which Minimoog is the real Minimoog? (It's like asking what's the best PAF clone in a pickup forum.)

You need to be feigning ignorance (disingenuity) in order to miss out on simple facts such as talking about the difference between an original painting vs. a print. Nowhere in the original painting are there CMYK dots or aliasing from discretization of the image.

You're changing the subject and setting up a straw-man by arguing that an average magazine reader can't detect the difference between a CMKY print of the original painting vs. a print of the print!

It takes a massive amount of knowledge and experience to become aware of the real differences and how to objectively measure them. This is like in my previous post where I'm pointing out there is little demand for such accuracy in emulations and that it is impractical to provide that accuracy even if there were sufficient demand.

I'm not talking about barely perceptible differences, I'm talking about quantifying major differences that you incorrectly attribute to different models which are in fact simply the result of a lack of any proper attempt at emulation of those properties at all.

Yep, now I'm pretty sure your criteria are much more specific and stringent than mine. I was definitely wrong, and I apologize.

[Jafo]

(Sorry, posting fsckup).

[foosnark]

As somebody who got heavy into hardware synths over the last year, my three cents:

1. There is no "impeccable" analog modeling but there's "pretty damned good, for most of what you're going to use this for" analog modeling.

1A. As musicians, we care about analog vs. digital. Unless your audience is very specific (Buchla owner's club or something), they don't. Many of them, in fact, have some pretty funny opinions about the subject because they don't actually know what you're talking about.

2. High audio rate modulation isn't a problem on hardware that's specced for it. But software plugins are forced to be CPU-friendly.

3. Nobody builds digital filters in hardware without a specific reason -- usually to do with cost. Code running on a chip you have to include anyway is cheaper than adding more circuitry, especially if your synth is polyphonic. Or maybe your module has 48 functions and very basic filter implementations are 3 of them. Much more rarely, it's something like a Z-Plane filter that embraces its digital nature.

[edit] 3A. There are some really cool things you can do with digital filters. ArcSyn does some of them. Resonators for modal / Karplus-Strong synthesis (in that territory between filter and delay) are generally much cleaner and better in digital implementations. But a lot (maybe too much?) effort goes into making digital filters imitate analog.

4. Generally, digital has won the oscillator war. Beyond basic Moog waveforms there's so much more diversity and power in digital oscillators. There are a few very good analog oscillators worth having, though.

5. In a modular system that's heavy on digital gear, analog signal and modulation paths do add a ton of character. I have no doubt about that. I just wouldn't use the word "warm" to describe it.

6. Anyone who insists on 100% analog everything, or 100% digital everything, is throwing away possibilities. I sympathize with in-the-box production, but you're still missing some cool stuff.

[Robert Randolph]

6. Anyone who insists on 100% analog everything, or 100% digital everything, is throwing away possibilities. I sympathize with in-the-box production, but you're still missing some cool stuff.

Everyone is always missing out on some cool stuff.

Someone that uses only digital will be privvy to cool things the 100% analog guy wont, and vice versa. It's always a balance.

[foosnark]

True enough. There's only so much time in the day.

[aciddose]

Yeesh, I wish I could keep up with you...

Fast typing and high detail are consequences of a number of things including personality traits, lack of anything more important to do (at least which I'd be willing to do) and 20 years of experience in programming

That said, be careful what you wish for.

Alas, it's not valid evidence, since it's not double-blind

Evidence of what?

That modern emulations are (was intended: not ?) good enough to reliably fool people.

Well yes, although it's important to keep in mind that when you start to try to define "which people?" this argument can break down significantly. I have no doubt that in general listeners don't care in the slightest. The most successful music today isn't entirely based upon which analog synthesizer was used, it's based upon the skill and (mostly) talent of the author as well as a significant element of "luck".

If we talk about other individuals however, such as those who have a fascination with the peculiarities of specific instruments, including analog synthesizers and their circuits (I call these people "synthists") that isn't just moving the goal-posts to another dimension but entirely redefining reality itself.

http://xhip.net/temp/another_dimension.mp3

The two groups are entirely different with no overlap whatsoever. A person can both be a music fan "listener" and have an interest in analog synthesizers "synthist" but the two need not conflict or overlap.

As a listener the more appropriate question would be "statistically, which instruments are the most popular?"

As an artist "statistically, which instruments are most likely to be used in the production of music which is most profitable?"

Both of these questions are technically unanswerable because time changes everything which brings us to the question: what is the point of asking the unanswerable?

Objective facts? Or subjective opinions? Nobody cares about subjective opinions in this discussion, otherwise we'd be arguing over how flavors of salad dressing or the wallpaper in your studio affect your mix.

What we're talking about is whether there are measurable differences between software emulations and the original hardware they claim to emulate.

Hmm... it occurs to me that maybe we're talking about slightly different things -- that my goals are lesser, easier to achieve.

Most definitely although I would say it is useful to consider all aspects of such topics as well as where they overlap. It also doesn't hurt to pose a hypothetical such as "... assuming we could objectively define the best synthesizer ... ?" and examine where it logically leads to.

As far as "tools you can use to satisfactorily produce music", plug-ins have long ago passed that hurdle for the most part. Only on the edge of that will you find people who aren't satisfied by software alone and they're more than happy enough to include a combination of hardware and software in their tools and equipment.

[Teksonik]

What moron would want to emulate Analog rubbish ?

[Terrafractyl]

It's not about analog vs digital anymore...that was 20 years ago. It's an archaic argument.

+1. We've been at the point where only confirmation bias, comb filtering, what one had for lunch and maybe phase of the Moon will find a difference.

Besides, any physicist well tell you that reality is digital anyway.

I love how you anti-analog guys always pop up to say this stuff here on KVR. I'm guessing you don't own a decent analog synthesizer!
Also you keep this myth going despite the fact that every Engineer worth his salt will tell you different... hell there is a plugin dev saying so just a few posts up.
There is a reason most plugin companies are still emulating analog stuff!
People that can't afford or justify analog gear will just always try and convince themselves that they don't need it. I know cause I used to be one of them!
As much as Uhe, NI, Synapse and all the rest can do a pretty awesome job at it, my little modular synth still blows them away at certain things.

Please note that nobody has said anything anti-analog -- just that digital has gotten to the point at which it emulates analog quite well enough, as per the thread title. Well enough that there is no reliable distinction which cannot be adequately explained by confirmation bias or random factors related to the recording or listening environment.

I wish it weren't so. I wish that analog meant something special that digital couldn't reproduce or replicate, but emulations have simply improved to the point that it fools human senses. Maybe if we were Vulcans or something, able to make finer just-noticeable distinctions.

Please point me to a proper double-blind experiment in which people consistently differentiate between an analog synthesizer and a really good emulation. Seriously, I really would love to be wrong about this. I would also love to live in a world in which dolphins had the intelligence of cats and stopped trying to rescue sinking pieces of wood. Or in which top engineers didn't spend thousands of dollars on a couple of pieces of wire in an analog summing box.

Hmm... that may have been a little too sarcastic. I'm sorry; I mean no offense.

Dude, i cant be bothered to find you a double blind test cause you clearly have no clue.

I never said you were anti analog, you just said digital was 'good enough' to get by and im sayng in some places its just not.
Analog vs digital is a very complex subject and i dont think your really seeing even part of the whole picture.
As far as synths and music I also agree with the above post in that some digital stuff can do plenty of things better than their analog counterparts . And visa versa!

As for great emulation, sure now some synths like the minimoog are pretty well covered for the most part.... But...
it doesnt take a double blind test to make a very simple patch with audio rate modulation and feedback on an analog synth, then try and do the same patch on something like reaktor or softube modular, and it is clear that these things are still very much not the same.
Everyones criteria are different, and i'm happy to say now that since modular synths became a thing i'm not so alone in wanting crazy fm sounds and wierd noises!

[aciddose]

Please point me to a proper double-blind experiment in which people consistently differentiate between an analog synthesizer and a really good emulation.
...

Dude, i cant be bothered to find you a double blind test cause you clearly have no clue.

I never said you were anti analog, you just said digital was 'good enough' to get by and im sayng in some places its just not.

...

As for great emulation, sure now some synths like the minimoog are pretty well covered for the most part.... But...

But ... I think that's exactly the only thing Jafo was really talking about. So it isn't so much disagreement as it is simply looking at things from a different perspective.

[funky lime]

[IncarnateX]

Remaining gab would be users who’d give a fvck so the devs would give a fvck instead of providing 13% approximations with photo realistic GUIs. But no way: Those who think they actually care have not heard a real analog classic beyond compressed Youtube vids and if they had they would probably be dissapointed. There is no magic, only sound characteristics that us old farts were primed to during the early days of electronic music and that is why we love them. Had digital synths been evented before analog and commercialized through decades of pop music, we would have longed for that sound instead. Lately I have been impressed by Syntronic for iOS. It is not a VA but a rompler and does not have “components modeled filters” but some average approximations. Yet those very fine sampled oscs take it a step beyond Diva, Legend, the Arturias and whatnot as far as analog realism goes. Have made me reconsider the whole deal: Drop the VAs. We cannot afford the CPU power to take them beyond their current state anyway. Make us some fine romplers instead with loads of features.

[Dasheesh]

The biggest wall is still the CPU which has a threshold, and likes to devote itself to running what windows wants instead of what you want, because you don't really own your PC or your instrument, they own you.

The biggest downside is the fact that digital can be very unreliable. It's constantly upgrading, so, what works today may very well not work tomorrow. That can be a problem for a professional player. The best way around it is to record all your sounds as stems and save them long term to work with later.

On the other side... everything as become a temporary wash anyway so what does it matter? Software sounds good as it is. It has gotten musical.