"Things that cannot be measured"

[dkinric]

I do believe that scientific instrumentation is capable of detecting sonic differences that are well below the threshhold of human "hearing". How our brains process those stimuli is fascinating, as is how our beliefs and personalities interpret our "perception" of sound. It's too bad we cannot measure highly variable neural "distortion" in our brains as easily as we can measure - and determine the audibility of - the distortion of signals in electro-mechanical systems.

Isn't this really the root of much of the disagreement in this hobby? Especially objective vs subjective stances. My interpretation of what the objectivist stance is, "this is what is coming out of this system" - That's all we are claiming, and that's all we can really do here on a mass level. How one's brain interprets those signals is too specific to the individual considering room, positioning and psychological factors.

The Subjective stance is equally valid in the final product - enjoying the sounds - but what your individual brain does with that information is not what is being presented here. We can only predict preferences based on statistical models. If you don't follow that model average, fine, but it doesn't change what your ears are receiving. It seems many that argue the subjectivist stance don't grasp this distinction.

 

[Xulonn]

Isn't it great to have an intelligent conversation about audio objectivists vs. subjectivists on an internet forum without it constantly degenerating into idiotic chaos and rancor?

While facilitating an excellent venue for intelligent written and video reviews and conversations, Amir has somehow managed to mostly avoid the nasty reactions against objectivists that exists on most other forums, and this seems to occur without blatant, heavy-handed moderation.

Of course, the very next post might weaken my argument...

 

[symphara]

Plain and simple there aren't any, everything we hear is a physical phenomena transferred via the air around us.

It is all physical phenomena but what we hear is in our brains and we cannot measure that. Not with any precision worth speaking of. Functional brain MRIs are an extraordinarily low-resolution instrument. The brain is simply far too complex, at approximately 1.5x10^14 synapses. Perhaps some AI in the distant future will be able to understand it, but I don't think we ever will.

Measuring the gear can be done pretty well, particularly for electronics which can be measured very accurately, speakers less so due to their analog, mechanical nature.

But believing that you can measure everything into the human brain perception is as foolish as spending 20k on audio cables.

So in this chain from artist performance to listener perception there's a lot that cannot be measured. It's probably good for electronics to be as transparent as possible, that's what I would base my buying decisions on, but I wouldn't be absolutely certain that a flat curve represents the maximum enjoyment in terms of perception.

We measure what we can, and I think it helps (a lot), but it's just hubris to think that we measure everything or that optimising for said measurements coincides with optimising for enjoyment.

 

[rgdawsonco]

No, you are correct that the processing that happens in human physiology is a "black box" in many ways. It is impossible to just measure things and, from that, predict what the perceived differences (or not) will be. Consider pitch again. There is NO WAY in the input signal to deduce or measure pitch - one has to incorporate a psychological model to measure pitch. What one can say is:

1) If there is no measurable difference (using good equipment) between two signals, then there can't be a psychological (perceivable/audible) difference.

2) If there is a measurable difference, the only way to know if it matters is if one has the correct (and currently often unknowable) transfer function from inputs to experience.

3) Given all of the above, maybe the best way to see whether a measurable difference matters is to do (correctly, as alluded to in a recent video by Amir) double blind comparisons.

4) What really matters is your experience. We are complex, transient beings. If it makes you happy, it makes you happy. So if your best listening experience is listening to something that happens to measure terribly, but for whatever reason you enjoy the sound of that signal, then it is the best for you. Case closed. Doesn't mean it is the best by any standard but your own, but that is all that actually matters. If you like box wine, enjoy box wine.

Very true. On the other hand, we humans sometimes like to accommodate (perceived) common or other's standards as well from a social perspective (e.g. impress the neighbors). For example, my wife likes wine. She would not serve boxed wine for guests, no matter how good, and no matter what her own standards were, because she would care what others might think/judge about the quality and how that may reflect on her. Humans can just be that way sometimes. But I tend to agree with you, If YOU like it, that is what matters most. I remember once a certain beer vender was advertising that it's beer (Brand A) was substantially preferred in blind taste tests. A friend claimed one day he preferred brand B. I ridiculed him, saying, "you only like it because of Brand B's advertising. I challenged him to do a blind test. He said, "Greg. I don't care why I like something, so go away." I think he made a good point.

 

[Classical EP]

It is all physical phenomena but what we hear is in our brains and we cannot measure that. Not with any precision worth speaking of. Functional brain MRIs are an extraordinarily low-resolution instrument. The brain is simply far too complex, at approximately 1.5x10^14 synapses. Perhaps some AI in the distant future will be able to understand it, but I don't think we ever will.

Measuring the gear can be done pretty well, particularly for electronics which can be measured very accurately, speakers less so due to their analog, mechanical nature.

But believing that you can measure everything into the human brain perception is as foolish as spending 20k on audio cables.

So in this chain from artist performance to listener perception there's a lot that cannot be measured. It's probably good for electronics to be as transparent as possible, that's what I would base my buying decisions on, but I wouldn't be absolutely certain that a flat curve represents the maximum enjoyment in terms of perception.

We measure what we can, and I think it helps (a lot), but it's just hubris to think that we measure everything or that optimising for said measurements coincides with optimising for enjoyment.

Exactly what I've been thinking. Different artifacts from the electronics and speakers may very well add to the perceived enjoyment, or may even trick the brain to feel the sound with the artifacts is more life like than without the artifacts.

 

[Jimbob54]

But I tend to agree with you, If YOU like it, that is what matters most. I remember once a certain beer vender was advertising that it's beer (Brand A) was substantially preferred in blind taste tests. A friend claimed one day he preferred brand B. I ridiculed him, saying, "you only like it because of Brand B's advertising. I challenged him to do a blind test. He said, "Greg. I don't care why I like something, so go away." I think he made a good point.

Except sometimes people confuse a refusal to sample something new/ different with having established preferences. Preference implies some experience of both sides of the coin. I think some refuse to try well measuring (maybe) cheaper kit because it doesnt have the prestige and backing of the audiophile media machine, whilst still declaring a "preference" for what they are used to.

The same applies in reverse of course. If one "prefers" low distortion well measuring electronics but refuses to even try something from the dark side to see if you might enjoy euphonic distortion then thats equally daft.

 

[mhardy6647]

Isn't FLAC lossless

Or did you mean lossy FLAC

Illogical... please explain...

 

[danadam]

Lossless FLAC?

Isn't FLAC lossless

From https://xiph.org/flac/format.html#architecture:

Similar to many audio coders, a FLAC encoder has the following stages:

• ...
• Prediction. The block is passed through a prediction stage where the encoder tries to find a mathematical description (usually an approximate one) of the signal. This description is typically much smaller than the raw signal itself. ...
• Residual coding. If the predictor does not describe the signal exactly, the difference between the original signal and the predicted signal (called the error or residual signal) must be coded losslessy. If the predictor is effective, the residual signal will require fewer bits per sample than the original signal. ...

 

[krabapple]

The answer is yes, FLAC is lossless.

 

[CMOT]

I think not. Pitch is the same thing as frequency (with a log scale, if you wish), and it is easily measured.

I don't want to get into a back and forth here. I will just assert the facts on pitch perception are very clear, see for example:

https://www.cns.nyu.edu/~david/courses/perception/lecturenotes/pitch/pitch.html
[Heeger is a world-renowned computational neuroscientist]

One's perception of pitch is not linear mapping of frequency. Things like virtual pitch must be incorporated into any system trying to infer pitch from sound signals. Incorporating something like virtual pitch is a psychoacoustic model that is only there because people work this way, not because of anything inherently in the signal.

See also:
Pitch perception: A dynamical-systems perspective
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC33127/

[PNAS is one of the most rigorous scientific journals in the world]

Some key lines from the abstract:

Because the auditory system is a complex and highly nonlinear dynamical system, it is natural to suppose that dynamical attractors may carry perceptual and functional meaning. Here we show that this idea, scarcely developed in current pitch models, can be successfully applied to pitch perception.

 

[mansr]

And yet, musical instruments are tuned according to a precise mathematical relationship, each note having a well-defined frequency, also called pitch.

 

[CMOT]

That is the pitch of the instrument, which is a combination of the frequencies produced by the particular instrument, but think about it - you could have a string instrument with a broken string, or a particular chord that omits the fundamental, yet you would still hear the same pitch. So tuning is just a way of adjusting the relative signals produced by that instrument, but it doesn't determine - nor is it linear - all of the pitches that instrument can produce. Since we are talking about human sound perception, we are presumably talking about human pitch perception. And that is more than the specific frequency combinations used for tuning - it is the space of all pitches we can perceive and what signals lead to those perceptions. But two different sets of signals can lead to the same pitch - this is only true because of how we combine the components of those signals in our heads.

 

[symphara]

No, you are correct that the processing that happens in human physiology is a "black box" in many ways. It is impossible to just measure things and, from that, predict what the perceived differences (or not) will be. Consider pitch again. There is NO WAY in the input signal to deduce or measure pitch - one has to incorporate a psychological model to measure pitch. What one can say is:

1) If there is no measurable difference (using good equipment) between two signals, then there can't be a psychological (perceivable/audible) difference.

2) If there is a measurable difference, the only way to know if it matters is if one has the correct (and currently often unknowable) transfer function from inputs to experience.

3) Given all of the above, maybe the best way to see whether a measurable difference matters is to do (correctly, as alluded to in a recent video by Amir) double blind comparisons.

4) What really matters is your experience. We are complex, transient beings. If it makes you happy, it makes you happy. So if your best listening experience is listening to something that happens to measure terribly, but for whatever reason you enjoy the sound of that signal, then it is the best for you. Case closed. Doesn't mean it is the best by any standard but your own, but that is all that actually matters. If you like box wine, enjoy box wine.

A thing about (1) is that it assumed that you have performed all possible measurements. That essentially you can produce a mathematical function that can perfectly predict the output on any given input of the device being measured.

I don’t think we‘re anywhere close to that on speakers. But on amplifiers? Can we measure everything about the behaviour of an amplifier?

 

[mansr]

That is the pitch of the instrument, which is a combination of the frequencies produced by the particular instrument, but think about it - you could have a string instrument with a broken string, or a particular chord that omits the fundamental, yet you would still hear the same pitch. So tuning is just a way of adjusting the relative signals produced by that instrument, but it doesn't determine - nor is it linear - all of the pitches that instrument can produce. Since we are talking about human sound perception, we are presumably talking about human pitch perception. And that is more than the specific frequency combinations used for tuning - it is the space of all pitches we can perceive and what signals lead to those perceptions. But two different sets of signals can lead to the same pitch - this is only true because of how we combine the components of those signals in our heads.

What you're saying is equivalent to insisting that measuring the length of a line is impossible because there exist optical illusions that trick us into misjudging it.

 

[CMOT]

A thing about (1) is that it assumed that you have performed all possible measurements. That essentially you can produce a mathematical function that can perfectly predict the output on any given input of the device being measured.

I don’t think we‘re anywhere close to that on speakers. But on amplifiers? Can we measure everything about the behaviour of an amplifier?

Right. It could be that our measurement devices are poor - too noisy, measuring the wrong thing, not sensitive enough, etc. so 1. is unlikely to always be limited to a guess as to what we should measure and how well we measure it.

Of course, all of this discussion about reproduction ignores the fact that we are listening to signals that are inherently design choices and data compression (independent of digital sampling/data compression) - the kinds of mics, the number of mics, and how the music was mixed - so our experience of listening to music through any system is going to be different from that of being at a live concert (particularly if unamplified). Maybe the closest one could get is to use a kemar (but not many recordings are made this way and you really need to listen to such recordings on the appropriate headphones).

 

[NTK]

... But on amplifiers? Can we measure everything about the behaviour of an amplifier?

How do the manufacturer make sure that the same "goodness" in the amplifier that got a glowing review from some reviewer is present in the one you bought?

Oh, they are manufactured "exactly the same way" with the "exactly the same materials". So that's it? What does that even mean? Does the phase of the moon at the time of manufacturing matter? How about which day of the week? Season of the year? No? Why? You have measured it? But you can't. That mystical quality is unmeasurable.

These are engineered products built using engineering principles based on physics. There is no escaping the fact that they are built to some measurable engineering specifications. Every transistor, capacitor, resistor, piece of wire (and/or how the piece of wire is constructed), coil, diaphragm, panel, magnet, etc. is constructed based on these principles. But the end result is unmeasurable but repeatable?

 

[Frank Dernie]

It is irrelevant how we sense sound.
Even if what goes on in one persons brain when they hear middle-C is the same as what goes on in another's when they see green it is totally irrelevant to sound reproduction.
All sound is is fluctuation in air pressure.
If a hifi can adequately reproduce the fluctuations in air pressure each individual will sense it in whatever way they psycoacoustically do and that side of it has no influence.
If we want to do direct neural injection of some sort we would need to understand that side of it, but not when exciting pressure fluctuations in the air.

 

[scott wurcer]

From https://xiph.org/flac/format.html#architecture:

If the predictor does not describe the signal exactly, the difference between the original signal and the predicted signal (called the error or residual signal) must be coded losslessy.

By exactly here they mean to the LSB so there is zero error. The predictor is some functional description and uses very little space. On decompression the predictor function + the residual error = exactly the input.

 

[scott wurcer]

We measure what we can, and I think it helps (a lot), but it's just hubris to think that we measure everything or that optimising for said measurements coincides with optimising for enjoyment.

Everyone's ear/brain is different, who ever said measurements could optimize for everyone? Measurements can only attempt to level the playing field, such as assuring that the sound field presented to each person is the same as possible. Individual preferences no matter where they come from are always there. I think in an audio forum this tread title is like waving a red flag at a bull. Things we can not measure does not include previously unknown physics that makes two cables sound "dramatically" different.

 

[StevenEleven]

One thing that I think about, which is in Toole’s book, is the adaptive change in perception of hearing a person experiences based on his or her many senses and cognition about the size and shape of a room. This is a very fluid and remarkable thing. I think perhaps there we are wandering into an area that is not fully measurable. Let me look for the page for those who have the third edition of Toole’s book. . . He first talks about it at page 17, section 1.7—“human adaptation, a reality that cannot be ignored.” He again refers to it at page 52, “Familiarity with the Room.” A logical inference I think you can make from what he is saying is that if you hear the same measurable phenomena in different rooms you will perceive it differently, based on other cues and thoughts your brain is receiving and creating with reference to the room.

How this hashes out in terms of whatever everyone is discussing here seems really messy so I am not even going to take a stab at it.

In terms of what @CMOT is saying about pitch or even arguably frequency being more perceptive than measurable at times, I think I am pretty sympathetic to that point of view. We have a fascinating and insightful thread (or at least I found it fascinating and insightful) on perfect pitch and relative pitch and the missing fundamental on this board somewhere. Edit—See: https://www.audiosciencereview.com/forum/index.php?threads/missing-fundamental.18930/

All of this with the caveat that I am not expert in any of this. But I can read and I can marginally play some music.

Speaking of which, I received an Apple Watch for my birthday and I was practicing piano and the watch told me I was being subjected to an occupational noise hazard! I was like, thanks a lot! So now if it’s just a matter of working out some fingering rather than being expressive I will slide on over five feet to the electronic keyboard. Indeed, my wife put a sound meter up by my ear while I was practicing piano and it easily got up to 90 to 95 decibels. I certainly did not perceive it as “loud” in that sense. If my A/V sound system that is only 20 feet or so away from my piano got that loud I would feel like I had to run out of the room (I get uncomfortable at 80 decibels or so). So there’s that too.