Can digital emulations be as good as analog hardware?

https://www.okmog.de/index.php/training/articles/can-digital-emulations-ever-be-as-good-as-analog-hardware

by olliqukr • 2 years ago

26 55 2 years ago

gsliepen• 2 years ago

Yes, because you only have to generate about 192 kilobytes/second to be able to make a pair of humans ear hear anything they could possibly hear, so the problem is bounded.

nayuki3• 2 years ago

Specifically this refers to 48 kHz, stereo, 16 bits per sample.

WanderPanda• 2 years ago

I found it fascinating how far we can go with even just < 10 kilobytes per second [1]

[1] https://bellard.org/tsac/

DaiPlusPlus• 2 years ago

Yes, because you only have to generate about 192 kilobytes/second to be able to make a pair of humans ear hear anything they could possibly hear, so the problem is bounded.

No FLAC file can replicate the subjective auditory experience of sticking a finger in your ear canal.

regularfry3• 2 years ago

You can generate a sound, but how do you know you generated the right sound? That's the point from the article that everyone seems to be missing. The problem isn't bounded because you can't know that your simulation is correct in finite time.

For that to actually matter you need to be simulating a more complex circuit than the one they give, but not by much: stick a feedback loop in and it gets funky.

teolandon• 2 years ago

It's very hard to take this article seriously when I'm flashbanged with one of the worst AI illustrations I've seen in my life right off the bat. Did the author even look at the image before publishing this?

Retr0id• 2 years ago

It's worse than one of those "holding the hot end of the soldering iron" stock photos.

st_goliath• 2 years ago

Yep, solder iron holds itself. Multi meter (pointing away from the user for some reason) displays fever dream nonsense.

With such tough competition, I guess https://www.hacker-stockphotos.com/ is out of business now?

bloopernova2• 2 years ago

Good lord you weren't wrong. How can anyone look at that image and think "yes, this is the professional look I want"?

Wherecombinator2• 2 years ago

Has there been any blind tests on analog purists? I have a few hardware synths myself but I have them for their tactility.

Only thing I find that hasn’t been successfully recreated in digital form is hi hats. They lack something bite or depth.

alfiedotwtf5• 2 years ago

If you’re talking 909, then there’s a reason why - the high hats on the 909 were samples lol :)

Wherecombinator2• 2 years ago

Nah talking 808 and 606s really but that said I still think the tr8s hats don’t match the original for whatever reason.

aj72• 2 years ago

Very skeptical of the thesis here. The practical limit is the sampling theorem and the bit resolution, not “computability.”

It is possible that there are persons who can resolve a 25th bit or be disturbed by aliasing associated with the sample rate chosen. But again, there is a bigger effect. It is well known that audiophiles like “warmth” which has often been attributed to analog saturation effects. But ultimately, what enters your ear is analog. It should be possible to design a digital system that reproduces this warmth at the analog ear, compensating for all harsh effects before detection.

amelius• 2 years ago

A big issue with digital audio production is latency.

frabert2• 2 years ago

At standard pressure and temperature, a sound wave travels at 343 m/s, IIRC. At a distance of 1m, those soundwaves take about 3ms to get from a pair of speakers to my ears. I can set the latency of my soundcard to as low as 0.5ms, which would mean an additional distance of ~18cm. I easily move my head that much when playing standing up. I don't see how that additional latency is the issue.

analog31• 2 years ago

For reference...

https://www.dmf.unisalento.it/~panareo/Application_note/AN-2...

quercusa• 2 years ago

Hard to imagine writing this article with zero references to Nyquist.

aj72• 2 years ago

Oh but I did. In the second sentence.

jerf• 2 years ago

There's another much, much larger issue, which is the complete lack of specification around the word "correct".

OK, so "a simple analog circuit like a passive 1-pole low-pass filter (see figure above) can have output values where digital emulations can't be sure if the result is correct". Sure, why not. But, stick another "passive 1-pole low-pass filter" next to it. Is it "correct"? Because it sure isn't "the same". That's the nature of analog circuitry like this.

"A person who has one analog circuit knows what the correct output is. A person who has two does not.", as the old saying about clocks goes.

The article is written from a mental perspective of there being a "correct", an exact, singular point that represents the one and only valid "correct" output for a digital simulation, and if the digital fails to capture that "correct" then it's just wrong. And it simply goes without question that if it is wrong obviously the music created with that wrongness has been destroyed or something, which is actually a pretty complicated topic of its own.

But "correctness" is actually a range in the world of analog... how are you going to prove that the digital simulation doesn't exist within the range of possibilities for a real circuit? Obviously there's plenty of "not even close" values. But digital getting to "close" can be hitting the bullseye because even if it doesn't match analog circuit A, it may well match analog circuit B with what is nominally the same design. (Which also suggests you can always expect that any digital simulation of a bit of analog gear you own will always not be quite exactly the same as what you have... but as long as it's in the range of the differences you can expect from another instance of your physical gear, which will also not be the same, it's basically "correct".)

Article also strikes me as written by someone who has at least some of the "Sampling fallacies and misconceptions" mentioned at https://people.xiph.org/~xiphmont/demo/neil-young.html . Doesn't come right out and say digital samples are jagged stairsteps but I think it's implied.

durpleDrank• 2 years ago

Man ... remember when Rebirth came out? lol that was fun

Having owned/used/borrowed ALL the greats and weird ones (Moogs, Junos, EMS, etc) I can safely say that "close enough" is good enough. I prefer not using a computer for music making these days because I already crunch 8 hour days debugging load balancer configurations ... I just can't give another 8 to the PC at the end of the day. It's so relaxing having a dedicated machine that is stand alone to tinker around with after dinner.

Let's face it, our dreams of becoming the next Trent Reznor isn't happening. If you're new, yeah use a PC and some Free VSTs. If you got the cash, sure go buy something. What's important is how YOU feel about it. I understand it can be fun comparing SERUM patches to a real ARP 2600 but in the end ... it pretty much does not matter and as long as something as 2 or 3 OSCILLATORS you can figure it out.

Does anyone really need anything more than Synth 1? https://www.kvraudio.com/product/synth1-by-daichi-laboratory...

If you read this far I'd highly recommend looking at these phone apps if you are new to music making.

KORG GADGET (surprisingly good) KOALA SAMPLER (incredible mpc style sampler)

alfiedotwtf5• 2 years ago

Exactly! Go for plugins UNLESS you’re chasing a particular sound and don’t want samples.

I’ve tried every 303 clone, and other bass synths, and until I got a RE-303, I was never truely happy. This thing IS what I was chasing all along and at half the price!

Everything else? Meh. A bass is a bass, a lead is a lead… but NOTHING can squelch like a real 303 if that’s what you’re gunning for!

lagniappe• 2 years ago

303 was THE sound of acid..

Ylpertnodi• 2 years ago

synth1

Don't forget the synth1 librarian (sorry, no link)! I managed to get three synth1's in my daw (plus the librarian). Thank you Mr Daichi.

kimi2• 2 years ago

Vince Clarke is using sw plugins nowadays. If it's good enough for Vince, it's good enough for me (says the man who plays with his Behringer 2600).

bloopernova2• 2 years ago

Any recommendations for recent tracks by Vince? I adore the Erasure album Chorus for the gorgeous sound Vince Clarke composed.

quijoteuniv• 2 years ago

Yes, nice to have standalone things. And anyway music does not care if we are snobs or not, is always benevolent

rahen• 2 years ago

Does anyone really need anything more than Synth 1? https://www.kvraudio.com/product/synth1-by-daichi-laboratory...

Any recommendation for Linux?

dr-smooth• 2 years ago

For all platforms, I recommend Vital (https://vital.audio/)

alfiedotwtf5• 2 years ago

Ctrl + f harmonics: 0 found Ctrl + f overtones: 0 found

When comparing painstakingly replicated plugins vs their real deal, I’ve found you can spot the difference if you know what to look for. Until harmonics, overtones, and even how each voice’s signal effects the heat and capacitance of the electronics, it can’t get it exactly…

Note that I didn’t say it can’t be good! There are a tonne of more than excellent emulations like the TAL-UNO-LX, MPC MiniD, and the Oberheim OB-X. Those sound magical!

We waste a lot of time and money chasing GAS (Gear Acquisition Syndrome) but deep down we all know that there’s nothing wrong with plugins these days

nayuki3• 2 years ago

This article is bullshit. It fails to mention the many limitations of analog hardware.

Analog signals are continuous waveforms that vary smoothly over time, capturing every nuance of the original sound.
The continuous nature of analog signals means they can theoretically capture an infinite amount of detail. When you play an analog record, for example, the sound you hear is a continuous representation of the original performance.

Analog signals get corrupted by noise everywhere. Noise of the read head. Wow-and-flutter in the speed variations of the phonograph or tape player. Noise of the wire. Crosstalk between wires. Noise of the amplifier. Noise of the mixer. Linearity and harmonic distortion. And so on. Digital systems can tightly control the noise and regenerate the signal perfectly.

digital signals are still approximations of the original analog waveform

And analog waveforms are approximations of the original analog sound wave.

These [analog] components interact with the signal in real-time, providing an infinite resolution

No, it's limited by noise and bandwidth. For example, an op-amp might only be good for 1 MHz.

For example, even a simple analog circuit like a passive 1-pole low-pass filter (see figure above) can have output values where digital emulations can't be sure if the result is correct. This non-computability presents a fundamental limitation in achieving perfect digital emulation.

They fail to elaborate on this. Maybe this is a reference to chaos theory and the sensitivity to initial conditions? But that wouldn't imply non-computability.

As technology evolves, the industry will continue to grapple with the balance between the precision of digital systems and the irreplaceable warmth and character of analog sound.

Don't tell me this is the same argument as "increasing density will change the character of the neighborhood"...

Anyway, the article also fails to mention ideas like:

* It is hard to control the variance of analog devices. This knob might behave noticeably differently from that knob, this transistor different from that one, and so on. If you have two analog guitar amps of the same model, it is legitimately possible for one to behave so differently from the other that you would hate to swap them.

* Digital algorithms are easier to design, implement, and debug. They are reproducible and modular. You can inspect any part of the signal processing chain without disrupting the result. This is not true in the analog domain, because for example, probing a circuit can change the behavior of the circuit.

* A must-see introductory video on digital sampling by Xiph.Org's Monty Montgomery: https://www.youtube.com/watch?v=cIQ9IXSUzuM [2013], https://xiph.org/video/vid2.shtml

regularfry3• 2 years ago

This totally misses the point. The article is trying to get at the fact that even if you try to emulate the analogue system including all those noise sources and bandwidth limitations, you still can't do it because of limitations on computation itself.

> For example, even a simple analog circuit like a passive 1-pole low-pass filter (see figure above) can have output values where digital emulations can't be sure if the result is correct. This non-computability presents a fundamental limitation in achieving perfect digital emulation.
They fail to elaborate on this. Maybe this is a reference to chaos theory and the sensitivity to initial conditions? But that wouldn't imply non-computability.

They have not failed to elaborate on it, you have failed to follow the reference they give. Let me help: https://www.researchgate.net/publication/360859184_On_non-de...

The argument for an RC circuit is basically that because it involves exponentiation, you can't calculate its exact output in finite time. And unfortunately this is somewhere that matters: as soon as you get feedback in the system or any other way for errors to accumulate over time, your simulated output is going to diverge from the theoretical. Whether that matters aesthetically isn't at issue. It's not identical, therefore it's not a match for the analogue system, and can't be.

Anyway, the article also fails to mention ideas like:

Neither of these matter in the context of the point the article is making. Variance within the emulation is trivial, and digital algorithms being easier to debug don't matter if the thing you're trying to emulate isn't emulable.

nayuki3• 2 years ago

even if you try to emulate the analogue system including all those noise sources and bandwidth limitations
Variance within the emulation is trivial

I pointed out those limitations not as a target for digital systems to emulate, but to refute the repeated insinuations that "Analog signals are continuous waveforms that vary smoothly over time, capturing every nuance of the original sound", "The continuous nature of analog signals means they can theoretically capture an infinite amount of detail".

because it involves exponentiation, you can't calculate its exact output in finite time

That hasn't stopped the exp() function from existing. https://en.cppreference.com/w/cpp/numeric/math/exp

as soon as you get feedback in the system or any other way for errors to accumulate over time, your simulated output is going to diverge from the theoretical

But then the same argument can be made against analog circuits. Two physical copies of a circuit necessarily cannot compute exp() exactly, exactly the same. So they will diverge too. So the digital version should be no worse than analog versions.

It's like you're arguing that a digital double pendulum (a well-known chaotic system) diverges from a theoretical mathematical version of it due to rounding and discretization errors. Like yeah, but any analog double pendulum is also an approximation of the mathematical ideal, and no two copies of the device will behave the same.

One difference is that it is relatively easy to increase or decrease the precision of a digital system - just increase the bit depth or sampling rate. Whereas increasing the precision of analog systems requires nontrivial engineering, and is sometimes impossible due to noise floors and other physical limitations.

schoen• 2 years ago

Thanks for the clarification. I'm still confused about the

Whether that matters aesthetically isn't at issue.

Presumably if I have a hardware random number generator that works by measuring thermal noise (directly or indirectly), I also can't emulate that in software. (If I could, cryptographic key generation would be insecure!)

But we know what the output of the hardware random number generator sounds like, subjectively, as audio: it sounds like white noise! (Or maybe some other kind of noise depending on the details of the digital to analog conversion process.)

So the idea that you can get an analog audio synthesis process that can't be emulated in software doesn't seem that surprising, but also potentially doesn't seem that interesting, except if someone also cares about something about how that process sounds subjectively to a human listener. But at that point, the theoretical argument about things that hardware can do that software can't emulate no longer applies directly anymore, right? That is, we can't extrapolate this theoretical argument about electronic circuits to something about qualitative human perceptions of the audio outputs of those circuits.

maxwell• 2 years ago

It's clearly slop.

Animats• 2 years ago

This article is bullshit. It fails to mention the many limitations of analog hardware.

Right.

The actual paper cited: [1]

That's an unrelated result. It comes from the observation that some functions are not usefully differentiable by sampling and differencing. If the function has fractal-like infinite detail, two samples taken close together have somewhat arbitrary differences. Some people who get wound up about the halting problem have latched onto this idea and try to make a big deal out of it.

Any analog system with non-zero noise, which is all of them, effectively has the same problem. You get a very noisy derivative if you try to generate a high-frequency derivative, because you're amplifying noise. This is why differentiators need a low-pass filter in front of them. This introduces some lag in a real time system. Aircraft vertical speed indicators are a classic analog device with this problem. A VSI is a barometric altimeter feeding a differentiator. They lag by several seconds. There are "instantaneous vertical speed indicators" with an accelerometer to provide some short-term correction.[2] Same problem, with a time constant of seconds, and in an analog device.

I ran into this problem in the motion interpolation of a virtual world, and fixed it.[3] A low-pass filter on velocity computed from position needed to be added. Completely different domain, same problem.

[1] https://arxiv.org/pdf/2205.12626

[2] https://www.mcico.com/resource-center/articles/instantaneous...

[2] https://community.secondlife.com/forums/topic/451190-merrily...

fuzzfactor• 2 years ago

If you have two analog guitar amps of the same model, it is legitimately possible for one to behave so differently from the other that you would hate to swap them.

This bug can actually be a feature even if it's just some random reason it sounds and plays remarkably better than the ones other players have :)

NickC25• 2 years ago

Can they be? Maybe.

Does the average end listener (the folks listening to your tracks) really give a damn if your lead sound or bass sound was made with a digital emulation of the ARP 2600 as opposed to the real thing (or the Korg re-issue)? No.

Can they tell the difference? Most likely not. Maybe if they've got a multi-million dollar listening setup comparable to what deadmau5 has and know exactly what to listen for (and how to tell the difference), but we know that's almost nobody. And even then, they most likely won't care if the actual track is good.

Bottom line - make good music with whatever tools you want to. If you can make it with a few plugins and a laptop, cool. If you can make it with fancy analog equipment, cool. The end result should be all that matters, not what shoes you wore on the journey.

nyrikki• 2 years ago

The end user is mostly irrelevant.

Lots of great songs were produced with rompler synths, and many of the killer electric guitars sounds people want to emulate were from people buying gear they could afford.

The minimoog is probably one of the most ubiquitous synths in popular music ever. I own one, and while it does have some qualities that VSTs can't reproduce, mostly it is how you interact with it that matters.

If as an artist you can connect to your tools, you can make great music.

While different than todays electronics, the E-mu Emulator was popular because it let you explore in a fairly intuitive way.

If you have ever worked with partials in a DX7 you will realize why so many songs used the presets.

People loved music even on 78rpm records, because it is all they had.

I like my EmU Procussion, because it brings back memories, none of which will ever be detectable on the end audio and modern daw samplers are far superior.

People buy expensive guitars and amps and justify it with a 'sound' but if you are creating music, it is just your connection to them that matters.

If you are trying to copy a sound, that is different.

glitchc• 2 years ago

I play guitar and own a Kemper profiler. It uses digital signal processing to emulate a variety of analog amplifiers. It is absolutely fantastic. Professionals agree and use it for live events. Profiling and modeling amplifiers put to bed the notion that analog is the only way to get a certain sound.

TheCleric• 2 years ago

Yeah I use a Line 6 Helix and I’ve had analog snobs surprised I was playing digitally. Is it 100% exactly like analog. Of course not. Is it close enough that >99.9% people would never notice? Yes.

gwbas1c2• 2 years ago

The continuous nature of analog signals means they can theoretically capture an infinite amount of detail.

Uhm, no: It's called frequency range and signal to noise ratio.

When you play an analog record, for example, the sound you hear is a continuous representation of the original performance.

Uhm, no: The roughness of a vinyl record is audible, pops and cracks are audible, and the last 1/3rd of the playable range of a 33.5rpm record can't even hold the highest frequencies that the human ear can hear. Often, deep bass needs to be filtered out so the needle doesn't skip out of the groove.

In contrast, digital signals are inherently limited by their sampling rate and bit depth.

These directly correlate to frequency response and signal to noise ratio.

There are important differences in details: It's very hard to make digital equipment "sound" analog, but that comes down to design choices. IE, tubes tend to handle clipping and overdriving differently than digital equipment. To claim that analog is infinite and continuous demonstrates a fundamental misunderstanding.

mikepurvis• 2 years ago

Indeed, and this should be trivially provable to anyone by the existence of 45RPM records for singles— if vinyl was truly "infinite" then why would you need to spin a record faster to get higher quality audio? Why not spin LPs even slower to get more content on them?

gwbas1c2• 2 years ago

Ironically, if you spin a record super-fast for higher quality audio, it will probably sound horrible. The holes often aren't perfectly in center; and it creates a waawaawaawaawaa (or wawawawawa on 45s) sound in any long sustained note.

Playing a record super-slow for a long playback would result in everything sounding off-key.

(This is why cd players make a shushushushush sound when you hold them up to your ear, it's the laser rapidly adjusting to an off-center hole.)

This also gets into an important difference between digital and analog: The ability to abstract away the physical playback mechanism from the representation allows for buffering so that playback is at the correct speed, even if the physical medium has an inconsistent speed. Digital also allows a continuous signal to be constructed from a broken signal: IE, streaming audio is constructed from many separate network packets.

In contrast, analog audio equipment is usually "instant," and digital audio equipment sometimes has a delay.

Venn1• 2 years ago

If you ever had to deal with recall sheets, the answer is yes. Being able to switch between sessions without having to readjust the settings for the entirety of the signal chain was a game changer.

https://www.barryrudolph.com/recall/sheets.html

Digital has the added benefit of sounding the same day in and day out. You can run audio through an analogue stack back to back and those recordings will not null.

For some, that's part of the charm of analogue hardware, but it's not something you want to deal with in production.

ChuckMcM• 2 years ago

I like this article, it asks an interesting question (albeit I think it is the wrong question but still it is an interesting one) about how accurately linear systems can be approximated by discrete systems. If you get into digital signal processing (DSP) either through audio/synth stuff or software defined radio or some other way, the trade-offs and how discrete mathematics differs from continuous mathematics are a very deep topic. If I understand the current papers on quantum mechanics it seems that physicists are thinking more and more of reality in discrete terms rather than continuous terms. All of which is fascinating, and yet the first question here for audio stuff should be;

What is the absolute best capability of the human ear to relative to its frequency response and dynamic range?

If you have an answer for that, then when your digital emulation has as good or better frequency response and dynamic range as the best ever measured human ear? Well that would be, by definition, indistinguishable from an analog counter part. (and yes, you can consider infra-sound, ultrasound, and blood bone harmonic resonance and all the other things the 'oxygen free' cable crowd sells you on :-)).

musictubes• 2 years ago

Yeah, this article isn't all that persuasive to me. The contrast between "analog warmth" and digital convenience doesn't make any sense to me. True emulation of vintage analog gear will always be an approximation since the originals were notoriously inconsistent. In any case the digital stuff sounds great and preferring the "warmth" of analog and its "technically infinite" amount of detail is little more than cork sniffing IMO.

One place that analog synths have an advantage over digital is with feedback patching. It isn't clear to me how VSTs or digital modules can cope with feedback like analog systems can. Feedback and cybernetic patching are pretty popular with Serge system in particular.

Pine_Mushroom• 2 years ago

I've dabbled in digital production since the late 90s and the state of the art is very impressive now compared to back then. U-he in particular really made some fine products, but there are many others too that I'm sure would pass blind tests against the originals they are emulating. I don't follow that world as closely anymore, but it seems there has been some stagnation in advancements over the past 5-10 years?

hilbert42• 2 years ago

Oh, this hory old debate again, when will it ever end?

I was brought up in the analog world but which one is better depends on the math. If one has sufficient bandwidth (Fourier components) and enough amplitude and phase linearity in one's ADC/DACs it makes no difference—any digital artifacts are outside any level of human perception.

Go even further and bats won't be able to distinguish the difference.

I prefer to trust the math any day rather than my subjective flaky ears.

That said, it's still never going to quell the argument, as the Golden Ear Brigade would have nothing left to talk about.

Can digital emulations be as good as analog hardware?

Related Stories