Any science to explain the “weight” or “3D depth” of hardware audio vs software that some people claim?

[DaveClark]

During the recent IK Multimedia 25th anniversary group buy, I was looking at the descriptions of the various tape machines and saw the following which reminded me of this thread:

“True Stereo: Even a perfectly aligned tape machine will have slight level, EQ and distortion differences between the left and right channels. This is good and part of the analog recording mojo! The T-RackS Tape Machine Collection perfectly recreates these subtle differences to deliver the real experience. If perfectly identical left and right channel processing is a must, simply disable this feature.”

Source: https://www.ikmultimedia.com/products/t ... collection
Source: T-RackS 5 User Manual, 2021, page 142.

A recent announcement on KVR Audio Forum / Effects regarding the release of four new tape machines contains:

“...while True Stereo operation provides a perfect recreation of the stereo image effect produced by subtle differences that inevitably exist between the left and right channels on an analog machine.”

Source: viewtopic.php?f=6&t=571792

This, based on what I believe was two years of characterization and development, is exactly the kind of thing that poster perpetual3 was describing before he was falsely accused of being unscientific and of violating some imaginary precept of forums which permits forced sharing of information. (*)


The bottom line is that IK Multimedia has produced software that has a switch which allows the user to go back and forth between the way in which software is usually written and an approximation of actual hardware that distinguishes between left and right channels. It’s been available for over two years.

The tape-machine stereo effect is rather subtle in my own testing, but it is definitely audible. It barely registers on the IK Multimedia meters but is visible. It’s perfectly believable that people would say that there is something about hardware that provides an enhanced perception of a sound in 3-D space compared to software which often neglects the subtle differences between right and left channels in hardware. If this one machine can cause such an audible effect, then it’s also believable that an entire chain of machines with these differences between left and right channels would provide an even more obvious enhanced perception of a sound in 3-D space compared to software that neglects these effects.

Recently I performed some SPICE simulations of two Fender Bandmaster guitar amps, one with half of the components shifted in alternating directions (up and down) while the other had half of the components shifted in the opposite direction (down and up). The amounts were 5% and 10% and resulted in a rather dramatic change in the audio image, especially for 10%. The intent here was to simulate two guitar amps coming off the factory line which had components which had a bimodal distribution due to hollowing out of a normal distribution for more discriminating customers. These results were not at all surprising.

Prior to this thread re-appearing, I had been building dual amp sims, which is why it caught my attention when it did reappear. I found that merely turning on a grid conduction model (based on work by Will Pirkle, Yamaha, and Peavey) in the first stage for one channel and the second stage for the other can create very wide stereo images as well as a very subtle widening. Hardly surprising.

Years ago while doing computational acoustics, I simulated coincident-microphone stereo recordings wherein the pressure gradient of the air in a room is resolved into two orthogonal or nearly orthogonal components. If the source and/or receiver are placed off-center in a rectangular parallelepiped, then a signal is created that, when played back through two speakers at a 90-degree or so angle with respect to the receiver, creates a pressure gradient that resembles the one which was present in the room. Here again, differences between left and right channels can produce stereo effects, effects which have nothing whatsoever to do with interaural time differences and which are often ignored in software. Neither surprising nor novel (Blumlein technique).

The point is that stereo perception is based not only on physiology and psychology, but also on the physics involved in a particular situation, i.e. by the physics of a room, the configuration of equipment, variations of component values, etc. resulting in a difference between right and left channels. To the extent that software ignores this, stereo perception can easily be less than that for sounds created and processed with real physical stuff.


(*) Regarding the false accusations mentioned above, some facts to consider:

In his 1905 paper creating special relativity, Einstein provided no references, disproving the claim that references must always be provided in science. This same paper began with “… ist bekannt.” “… is known.” Similarly, the 1925 paper by Heisenberg which founded quantum mechanics began with “Bekanntlich lässt sich ….” “It is well-known …” disproving that all claims in science need to be supported with evidence provided by the claimant. This is not necessary for well-established claims, nor is it necessary for informal publications and discussions. Audiences, even for formal papers, are expected to have a certain minimum level of education. Authors and presenters are not obligated to provide remedial training. A forum is defined as a meeting place for “open discussion,” not as a meeting place for “sharing knowledge.” A forum is not defined as a commune. Physics presumes all of mathematics which is neither an unproven claim nor a collection of unproven claims, disproving that a presumption is an unproven claim. In theory of argument, it is the position that would prevail lacking controversy. Theory of argument applies to all discussions which include controversy, including both legal and scientific debate. A premise is a basis for future argument, while a conclusion is an opinion or judgment after investigation or thought. For example, the following two paragraphs contain, from the perspective of this particular post, conclusions, not premises.

The false accusations were based on 1) an overly simplistic, mythological view of what science is that is completely at odds with the actual history of science and 2) an incorrect view of what a forum is.

In addition, 3) there is a serious lack of knowledge in most forums about the ways that have been developed for resolving controversy over the past 2000+ years, and this ignorance causes all sorts of unnecessary problems.

In the hopes that it will help, according to theory of argument, aka “rhetoric,” aka “informal logic”: If there is a controversy involving A and B where A has the presumption, A doesn’t have to prove anything, regardless of how much B whines and stomps his/her feet, or attempts to shame and ridicule A. Not until B provides a valid counterargument, evidence to the contrary and so on is A obligated to respond to B (burden of rejoinder). perpetual3’s detractors provided neither valid counterarguments that addressed the actual issue at hand nor evidence. Until they do, nothing more needs to be said.

[Christian Schüler]

These topics tend to surface a lot. What is the magic of real (TM) hardware?

I have a theory: Hardware devices tend to sound better than software programs (even if the hardware is driven largely by software internally), because more and better craftsmanship goes into making hardware devices.

Software can be released early and often. Therefore, more often than not, it is released as soon as it gets 'good enough'. You can always update later.

For hardware, on the other hand, it pays to have engineers think hard about the implementation and tweak every last bit of the algorithms before going into production, because that cost is amortized over thousands of devices and is dwarfed by the production cost of the devices themselves. Making hardware forces you to think more carefully about it.

[earlevel]

These topics tend to surface a lot. What is the magic of real (TM) hardware?

I have a theory: Hardware devices tend to sound better than software programs (even if the hardware is driven largely by software internally), because more and better craftsmanship goes into making hardware devices.

Software can be released early and often. Therefore, more often than not, it is released as soon as it gets 'good enough'. You can always update later.

For hardware, on the other hand, it pays to have engineers think hard about the implementation and tweak every last bit of the algorithms before going into production, because that cost is amortized over thousands of devices and is dwarfed by the production cost of the devices themselves. Making hardware forces you to think more carefully about it.

It would be better if we compared specific products, but in this same general sense I'll float a counter argument (as an electrical engineer working on audio circuits that eventually specialized in DSP):

First, "software is released early and often"—I don't see any fewer man hours being put into software than hardware. And, software gets improved far more often than hardware, so the total design hours can become huge over time, compared with hardware. Design a hardware EQ, it likely stays that way for the rest of its product life.

Also, "more and better craftsmanship goes into making hardware devices": I can think of some cases why you might say this, certainly Rupert Neve was a craftsman who designed the same things using multiple different approaches during his lifetime, all excellent. But you have to assume that there are not digital designers that have done the same, in order to think that's unusual to analog hardware.

If you look at the bulk of prized audio designs—EQs, including Pultec, for instance, or those in consoles, or similarly prized compressors like LA-2A, 1176, these are not complex designs, they were designs that worked. Different designs can work too—LA-2A and 1176 are mostly trying to solve the same need, but do it in different ways and have different strengths and shortcomings. And there were countless other designs for that need that failed and we don't even know their names.

And that's what I think it boils down to. For many audio designs, analog is not perfect to our ears, different solutions will impart different sonic finger prints to the audio. Over the course of decades, the ones that left favorable imprints are the ones we prized and kept.

In some cases, it's not just the sonic signature, but usability—if you have two hardware EQs with limited adjustability that sound just as good, and one gave its few frequency settings in spots that happen to work out really well for mixing rock and roll, that's the one that survives the test of time. Now, make a plugin EQ that sounds just as good, but gives you the ability to sweep to any frequency, and gives you five ways to define Q. For some, that might take it from twisting a knob on a console to select one of three or four settings—"ah, the third one sounds great on the kick, I think I'll reach for that first every time I'm in a similar situation"—and moving on with satisfaction, to always hunting for something on the plugin EQ.

Anyway, that's my story and I'm sticking to it—the analog hardware worked itself out over decades, and we largely recycle those designs for new gear still. In digital, an effect can sound like anything, but we most often try to make it sound like something we already know does the job. Because if Pultecs sounded perfect, we wouldn't use them. But that's what life is about anyway—we can make walls that are perfectly flat and smooth, but prize hand-plastered walls as craftsmanship.

[BONES]

I have a theory: Hardware devices tend to sound better than software programs (even if the hardware is driven largely by software internally), because more and better craftsmanship goes into making hardware devices.

That's provably untrue. You just have to look at how bad some hardware companion software is to see that hardware vendors are at least as capable of just slapping it together as anyone.

Software can be released early and often. Therefore, more often than not, it is released as soon as it gets 'good enough'. You can always update later.

Not in my experience as a beta-tester. OTOH, my analogue Uno Synth has had updates that add/change functionality such that what's written on the front panel now no longer reflects how the instrument works.

Making hardware forces you to think more carefully about it.

I'd suggest it restricts your thinking because everything you want to implement likely has a cost.

[zerocrossing]

These topics tend to surface a lot. What is the magic of real (TM) hardware?

I have a theory: Hardware devices tend to sound better than software programs (even if the hardware is driven largely by software internally), because more and better craftsmanship goes into making hardware devices.

Software can be released early and often. Therefore, more often than not, it is released as soon as it gets 'good enough'. You can always update later.

For hardware, on the other hand, it pays to have engineers think hard about the implementation and tweak every last bit of the algorithms before going into production, because that cost is amortized over thousands of devices and is dwarfed by the production cost of the devices themselves. Making hardware forces you to think more carefully about it.

This was a partially correct answer in the early 00s, but that situation is no longer. Software has really matured. Partially due to it’s own maturation, but also due to the processing becoming so good that shortcuts no longer need to be taken to the degree that they once were.

I now no longer think that any hardware is better than good software. Is it the same? No, but I have plenty of both and from my vantage point it becomes more about picking something that inspires you sonically and had the feature set that makes you happy. What I will say is, that I think analog is still the best at being analog and digital is still the best at being digital, and by that I mean that there is really little reason to buy all digital hardware synths unless it’s for the interface, which is definitely not nothing. That said, analog emulation isn’t something I think of as an ugly duckling. It is its own thing and has capabilities and a sound of its own.

[zerocrossing]

Making hardware forces you to think more carefully about it.

I'd suggest it restricts your thinking because everything you want to implement likely has a cost.

This. If you look at the overall synth landscape, it’s clear that you have three sectors. Fixed architecture hardware, modular and software. Modular and software generally have much more varied and interesting offerings than fixed architecture synths. I suspect it’s because of what you said. The only “price” the developer has to pay to implement a feature is the time it takes to design it. In hardware, unless it’s something that can be done in firmware, the developer has to factor in the cost of the parts and manufacturing.

Modular is closer to software in that it does seem to have a very innovative market, but it’s much more of a niche market. I’m not sure what the economics are, but my guess is that the lack of market size is overcome by the fact that the entire market is hungry for innovative features. The cost is very high too. I once thought I’d replace my Pro 2 with a modular setup, and to just basically do what the Pro 2 does would have bumped up the cost by a factor of four, if memory serves.

[HAL76]

During the recent IK Multimedia 25th anniversary group buy, I was looking at the descriptions of the various tape machines and saw the following which reminded me of this thread:

“True Stereo: Even a perfectly aligned tape machine will have slight level, EQ and distortion differences between the left and right channels. This is good and part of the analog recording mojo! The T-RackS Tape Machine Collection perfectly recreates these subtle differences to deliver the real experience. If perfectly identical left and right channel processing is a must, simply disable this feature.”

Source: https://www.ikmultimedia.com/products/t ... collection
Source: T-RackS 5 User Manual, 2021, page 142.

A recent announcement on KVR Audio Forum / Effects regarding the release of four new tape machines contains:

“...while True Stereo operation provides a perfect recreation of the stereo image effect produced by subtle differences that inevitably exist between the left and right channels on an analog machine.”

Source: viewtopic.php?f=6&t=571792

This, based on what I believe was two years of characterization and development, is exactly the kind of thing that poster perpetual3 was describing before he was falsely accused of being unscientific and of violating some imaginary precept of forums which permits forced sharing of information. (*)


The bottom line is that IK Multimedia has produced software that has a switch which allows the user to go back and forth between the way in which software is usually written and an approximation of actual hardware that distinguishes between left and right channels. It’s been available for over two years.

The tape-machine stereo effect is rather subtle in my own testing, but it is definitely audible. It barely registers on the IK Multimedia meters but is visible. It’s perfectly believable that people would say that there is something about hardware that provides an enhanced perception of a sound in 3-D space compared to software which often neglects the subtle differences between right and left channels in hardware. If this one machine can cause such an audible effect, then it’s also believable that an entire chain of machines with these differences between left and right channels would provide an even more obvious enhanced perception of a sound in 3-D space compared to software that neglects these effects.

Recently I performed some SPICE simulations of two Fender Bandmaster guitar amps, one with half of the components shifted in alternating directions (up and down) while the other had half of the components shifted in the opposite direction (down and up). The amounts were 5% and 10% and resulted in a rather dramatic change in the audio image, especially for 10%. The intent here was to simulate two guitar amps coming off the factory line which had components which had a bimodal distribution due to hollowing out of a normal distribution for more discriminating customers. These results were not at all surprising.

Prior to this thread re-appearing, I had been building dual amp sims, which is why it caught my attention when it did reappear. I found that merely turning on a grid conduction model (based on work by Will Pirkle, Yamaha, and Peavey) in the first stage for one channel and the second stage for the other can create very wide stereo images as well as a very subtle widening. Hardly surprising.

Years ago while doing computational acoustics, I simulated coincident-microphone stereo recordings wherein the pressure gradient of the air in a room is resolved into two orthogonal or nearly orthogonal components. If the source and/or receiver are placed off-center in a rectangular parallelepiped, then a signal is created that, when played back through two speakers at a 90-degree or so angle with respect to the receiver, creates a pressure gradient that resembles the one which was present in the room. Here again, differences between left and right channels can produce stereo effects, effects which have nothing whatsoever to do with interaural time differences and which are often ignored in software. Neither surprising nor novel (Blumlein technique).

The point is that stereo perception is based not only on physiology and psychology, but also on the physics involved in a particular situation, i.e. by the physics of a room, the configuration of equipment, variations of component values, etc. resulting in a difference between right and left channels. To the extent that software ignores this, stereo perception can easily be less than that for sounds created and processed with real physical stuff.


(*) Regarding the false accusations mentioned above, some facts to consider:

In his 1905 paper creating special relativity, Einstein provided no references, disproving the claim that references must always be provided in science. This same paper began with “… ist bekannt.” “… is known.” Similarly, the 1925 paper by Heisenberg which founded quantum mechanics began with “Bekanntlich lässt sich ….” “It is well-known …” disproving that all claims in science need to be supported with evidence provided by the claimant. This is not necessary for well-established claims, nor is it necessary for informal publications and discussions. Audiences, even for formal papers, are expected to have a certain minimum level of education. Authors and presenters are not obligated to provide remedial training. A forum is defined as a meeting place for “open discussion,” not as a meeting place for “sharing knowledge.” A forum is not defined as a commune. Physics presumes all of mathematics which is neither an unproven claim nor a collection of unproven claims, disproving that a presumption is an unproven claim. In theory of argument, it is the position that would prevail lacking controversy. Theory of argument applies to all discussions which include controversy, including both legal and scientific debate. A premise is a basis for future argument, while a conclusion is an opinion or judgment after investigation or thought. For example, the following two paragraphs contain, from the perspective of this particular post, conclusions, not premises.

The false accusations were based on 1) an overly simplistic, mythological view of what science is that is completely at odds with the actual history of science and 2) an incorrect view of what a forum is.

In addition, 3) there is a serious lack of knowledge in most forums about the ways that have been developed for resolving controversy over the past 2000+ years, and this ignorance causes all sorts of unnecessary problems.

In the hopes that it will help, according to theory of argument, aka “rhetoric,” aka “informal logic”: If there is a controversy involving A and B where A has the presumption, A doesn’t have to prove anything, regardless of how much B whines and stomps his/her feet, or attempts to shame and ridicule A. Not until B provides a valid counterargument, evidence to the contrary and so on is A obligated to respond to B (burden of rejoinder). perpetual3’s detractors provided neither valid counterarguments that addressed the actual issue at hand nor evidence. Until they do, nothing more needs to be said.

There are some people around here who want to write differences away and make it look as if most people thought that there were none anymore. May sound like a conspiracy theory, but you might identify some actors with "zero tolerance" yourself and also ask "cui bono" before you believe philosophical assessments on the not existing overall (!) differences. ROFL.

[BONES]

In his 1905 paper creating special relativity, Einstein provided no references, disproving the claim that references must always be provided in science. This same paper began with “… ist bekannt.” “… is known.” Similarly, the 1925 paper by Heisenberg which founded quantum mechanics began with “Bekanntlich lässt sich ….” “It is well-known …” disproving that all claims in science need to be supported with evidence provided by the claimant.

Don't be absurd. Einstein posited a theory, he was not expressing a law, like Newton's laws of motion. And almost 100 years later, we still refer to much of Einstein's work as "theory", not as "law". It's really good theory, in that it has predicted many natural phenomena that we have subsequently confirmed by experimentation and observation, like Gravity Waves, but it's still considered a theory. And Heisenberg had an experiment to prove the Uncertainty Principle, which is why it's a principle, not a theory.

[jamcat]

Y’all’s some nerds up in here

[melomood]

and not one stopping to think claims in science not supported with evidence can turn in to scientific claims which are eventually debunked

Me,I prefer software because the sound weighs less,it's sleek and can pull off a simple black evening dress. Also you can fit more in a car with the extra headroom

[DaveClark]

In his 1905 paper creating special relativity, Einstein provided no references, disproving the claim that references must always be provided in science. This same paper began with “… ist bekannt.” “… is known.” Similarly, the 1925 paper by Heisenberg which founded quantum mechanics began with “Bekanntlich lässt sich ….” “It is well-known …” disproving that all claims in science need to be supported with evidence provided by the claimant.

Don't be absurd. Einstein posited a theory, he was not expressing a law, like Newton's laws of motion. And almost 100 years later, we still refer to much of Einstein's work as "theory", not as "law". It's really good theory, in that it has predicted many natural phenomena that we have subsequently confirmed by experimentation and observation, like Gravity Waves, but it's still considered a theory. And Heisenberg had an experiment to prove the Uncertainty Principle, which is why it's a principle, not a theory.

These are not asburdities; they are facts.

Einstein published a very highly regarded scientific paper that had no references. This is contrary to what posters had said previously about references always being required to be provided by the claimant about all claims they make in science. Your post misrepresents Einstein's relativity paper which, in addition to introducing new theory based on two principles that could be regarded as laws, also includes claims about results from experiments.

Heisenberg's paper had references, but as I specifically showed, not for all claims made in the paper. The latter is also contrary to what other posters said about references being required from the claimant about all claims made.

[mystran]

I think we're missing a degree of nuance here. It is one thing to require a reference for everything (and that would be incredibly silly; you have to draw the line on "well known" somewhere), yet it is another matter entirely to expect someone to be able to defend their claim when that claim is disputed. In a sense, it's a difference between references up front vs. references when requested. I'm certain Einstein would have been able to provide references upon request.

Also to clarify, I'm personally perfectly happy to consider it "well known" that two nominally identical analog circuits processing two channels of audio in parallel won't ever produce the exact same results. Yet, the question of "weight" or "3D depth" is really a psycho-acoustic question and I would genuinely like someone to provide references any high-quality studies on the subject. I have tried searching for them, but so far I have no been successful. Variation in stereo imaging does seem like one plausible cause, but are some types of processing differences more important than others and what would explain "3D depth" in a single-channel monophonic analog synthesizer? Also how strong is the contribution of the placebo effect on this topic?

[ghettosynth]

and not one stopping to think claims in science not supported with evidence can turn in to scientific claims which are eventually debunked

Me,I prefer software because the sound weighs less,it's sleek and can pull off a simple black evening dress. Also you can fit more in a car with the extra headroom

This is correct. I should point out though, that with the advent of chiplets, which can be thought of as an additional dimension. I can believe that such 3d hardware is more capable of producing 3d sounds.

In fact, we can probably trace this all they way back to hand wiring which really does have more dimension. This is why older devices with through hole parts sound better and really old synths that had their voice boards placed vertically sound the best. There's a reason that you can't get your hands on a Voyetra 8 today.

Things got really flat in the 90s and this would explain why the gear from that era is so dimensionless in its sound.

I hope that I've cleared things up for everyone.

[melomood]

I just checked out a Voyetra 8. That baby looks like it has weight and depth

[ghettosynth]

I just checked out a Voyetra 8. That baby looks like it has weight and depth

Right!?! Mystery solved. We should go back to talking about cats now.