How The Ear Works

[garylkoh]

The problem with science is that it not infallible and it is not exhaustive. If it were, scientists would know everything and we would still be eating margarine because it contains no cholesterol and hence it is good for you. That is, until scientists figured recently that hydrogenating vegetable oils created trans-fatty acids which are worse for health than cholesterol.

Even in our own hobby, when scientists discovered how to measure total harmonic distortion, and designers figured that putting in lots of negative feedback could reduce this measurement, we had wonderfully specified amplifiers that did not sound like music. Or when scientists in their research concluded that by all means of measurement the CD was perfect sound forever. If it had not been for the people who trusted their ears more than their measurements, where would we be today?

Research is still on-going on how we hear. Before the ears otoacoustic emissions (OAE) were discovered in the late 70's, scientists did not understand how the ear could have a dynamic range of 140dB. And it took over 20 years before otoacoustic emissions were generally accepted because then the instruments were sensitive enough and had a low enough noise floor to measure it reliably.

Now, scientists are exploring the cochlear amplifier - that the ear has an active, non-deterministic, non-linear amplifier. They still can't understand much of the workings - the cochlear (as far as has been so far determined) does not simply amplify what's coming in, it changes OAE in a non-deterministic way to interfere with the incoming soundwaves and that results in a non-linear amplification. Scientists believe that this improves selectivity (understanding speech in the presence of masking noise) more than it improves sensitivity.

Research is still on-going, and more questions are still being raised.

There are two types of scientific minds - the ones that are half full, and until you can disprove what they know will continue to be half full. There is the other type that is half empty, and will continue to be always half empty because there is still much out there to learn. (Then there are the full minds and the empty ones, but that's off topic if we are talking about science.)

This is why my research is not into the science of measurement and specifications, but in the science of hearing - if we do not yet know how we hear, how do we know that our suite of specifications and measurements is already complete? The more I find out, the more I find that I don't know. But some of that research has already benefited the people who believe.

Can I measure it? No, I don't know how - it's based on how we hear, not how we measure. Can I prove it? No, except by listening, and there are those here who will say that listening is not good enough, you need DBT. I once said to a friend (please forgive me if you find this insulting because he did) - show me a double-blind test for love, and I'll show you a double blind test for enjoying music. If the very nature of the test changes the test (quantum entanglement paradox) then the test is flawed.

Hearing at the threshold of perception is all about the mind - if you can convince the mind that there is danger, the sensitivity of your hearing improves.

I not only believe, but I live by a philosophy - Discovery consists of seeing what everybody has seen, and thinking what nobody has thought. So, I will continue on this quest to find out what it is that we don't know exists but some of us believe exists, and will continue to ignore all exhortations to prove it exists using standards and measurements that I don't know is complete and comprehensive.

 

[Gedlee]

Both science and listening are required, neither alone is of much use. It's a loop that must be closed. We must verify what we (think) we hear with measurements and we must verrify what we (hope to)measure with subjective studies. Either one without the other is a problem only half done.

Your reflections on science are quite negative and I don't share your opinions. Consider this; there is a theory of everything that is currently being tested and if found validated by experiment will mean that we have found the theory of all that there is in the universe - the Unified Field Theory. And you think that science is somehow lagging? Your examples of the ear and hearing are quite old and all this work has been completed for a decade or more. We know precisely how our hearing works, its not a mystery. You may not be aware of these results, but they are known. The ear is "active", there is a cochlear amplifier, it is the reason that we have such high selectivity and sensitivity. Science in audio is why it has adavnced despite the audiophiles that try to hold it back with false beliefs unproven beliefs that defy science. As I said, only when measurements and subjective evaluations are in complete agreement can we say that we know something. Statements based on one or the other as sole support are not worth arguing about. But thats not to say that we can't hypothesize about relationships until the results are 100% in agreement, because we can and should. But the goal must always be to bring the two things ever closer into agreement and to use all available data to refine our models. One must also weigh the data by its typical likeihood of stability and accuracy. Subjective data is notoriously unstable as anyone who has done these kinds of experiments will tell you. Measurements have been shown to be stable, accurate and precise, but often measure things which are not relavent and fail to show any subjective correlations. Both sides have their uses and their problems and it is wise to know what they are. Belittling science is certainly not what I would do - its certainly not what I do do.

 

[amirm]

Earl, I agree the science of how we hear is good and quite exhaustive as that has medical benefits that researchers are interested in. How we sense fidelity though, is extremely young and primitive as there is no medical benefit in that.

In my previous job, we were designing psychoacoustic models for audio compression. We developed quite a few and I have to tell you that it was very difficult to pick between them. They would all get the first order effect right but not second and third. Some would sound better on this track, but not another while the ear of a trained listener had no trouble producing consistent evaluations.

Concepts like which set of distortion is better than another set present an infinitely complex set of scenarios that are next to impossible ascertain. Is 5 ns of jitter at 10 Khz worse or better than 7 ns of jitter at 8 KHz? Sure, I can tell you based on psychoacoustic modeling that jitter in low frequencies don't matter (first order effect). But cannot answer the example just given as my crude models don't scale. Throw in the attributes of the jitter in there and it becomes a 3-dimensional problem which is incredibly complex to get your arms around.

If I had a perfect model of fidelity, I could write software to implement it and do away with all listening tests. The fact that we still rely on hearing tests is sure way of knowing we are not there.

Audiophile world is not about first order effects. We are fortunate enough that outside of speakers, we are pretty good there. Frequency response of a CD player is excellent. Its noise level very low. We know we can hear rolled off high frequencies and noise at some levels. With those taken care of, we are now into second order effects that are harder to quantify. Audiophiles of course push even further into third-order and higher as they talk about esoteric tweaks.

So while I feel great about how much science we have, I am humbled by how much we don't know. I read your papers and appreciated the understanding you bring to your field. You are proud of your discoveries there. That tells me that others do not know what you know and hence, it can't be the case the science of fidelity is known or there would be no discovery.

 

[garylkoh]

Earl, if you say that all the work on the ear and hearing has already been completed for a decade or more then there is nothing more for me to say.

 

[Gedlee]

Amir

Its not about "having all the answers" as you suggest, its about "using what we know". You are quite correct that audiophiles worry about third, fourth order effects, but then they seemingly ignoring the first order ones. People are only now starting to appreciate the critical need for Constant Directivity in a loudspeaker (even Linkwitz talks about it now!) but its not new at all and its clearly a highly audible, anyone can hear it, first order effect. Thats what really gets me is peoples obsession with things that "maybe" someone could hear, but ignore effects that everyone can hear because they don't understand them or they aren't easy to correct. Audiophiles appear to obsess over those things that they can readily change while completely ignoring those things that are difficult even if they do completely dominate the sound quality. Compared to the room and the interaction of the speakers with the room, LF modal effects, reflection patterns, etc. all else pales by comparison.

 

[amirm]

Amir

Its not about "having all the answers" as you suggest, its about "using what we know". You are quite correct that audiophiles worry about third, fourth order effects, but then they seemingly ignoring the first order ones. People are only now starting to appreciate the critical need for Constant Directivity in a loudspeaker (even Linkwitz talks about it now!) but its not new at all and its clearly a highly audible, anyone can hear it, first order effect. Thats what really gets me is peoples obsession with things that "maybe" someone could hear, but ignore effects that everyone can hear because they don't understand them or they aren't easy to correct. Audiophiles appear to obsess over those things that they can readily change while completely ignoring those things that are difficult even if they do completely dominate the sound quality. Compared to the room and the interaction of the speakers with the room, LF modal effects, reflection patterns, etc. all else pales by comparison.

At the general level, I have no disagreement with what you are saying. But let's drill down. Take Steve's room. He has $30K in room acoustics, and extreme amount of money spent on speakers. I know they are not your brand but let's assume for the sake of discussion that he has achieved the pinnacle there. Should he stop there and not search for more he can do? Many of our vocal audiophiles in this forum fall in that category. They have achieved first and second order nirvana and are then searching for more as this is their hobby.

For Steve, that came in the form of R2R tape format and tube amps. This is where the world gets complicated. We could say that he is going backward in the first order parameters with both of these. I have to admit that I am stomped here. I have heard his system and it sounds wonderful even though the specs are poor. Low wattage amps. Lots of distortion. The simplest explanation is that both of us are fools and that we are detecting preference rather than fidelity. That could be right. But just as well it could be that there is something there with respect to fidelity which we still don't quite understand .

 

[Ron Party]

At the general level, I have no disagreement with what you are saying. But let's drill down. Take Steve's room. He has $30K in room acoustics, and extreme amount of money spent on speakers. I know they are not your brand but let's assume for the sake of discussion that he has achieved the pinnacle there. Should he stop there and not search for more he can do? Many of our vocal audiophiles in this forum fall in that category. They have achieved first and second order nirvana and are then searching for more as this is their hobby.

For Steve, that came in the form of R2R tape format and tube amps. This is where the world gets complicated. We could say that he is going backward in the first order parameters with both of these. I have to admit that I am stomped here. I have heard his system and it sounds wonderful even though the specs are poor. Low wattage amps. Lots of distortion. The simplest explanation is that both of us are fools and that we are detecting preference rather than fidelity. That could be right. But just as well it could be that there is something there with respect to fidelity which we still don't quite understand .

Your hypothetical assumes facts not in evidence. As such it is a strawman argument.

 

[amirm]

Your hypothetical assumes facts not in evidence. As such it is a strawman argument.

??? Not sure what your answer means Ron. I was responding to Earl saying that first order effects like the room and speakers are the things people should worry about. I answered that once people get over that hump, what next? Should they give up and not pursue perfection in second and third-order levels? What in there is not factual? Steve's room?

I am not asking you to believe that tube amps and R2R provide better fidelity. Read the last part where I said that may or may not be true. So if that is what you are objecting to, then you are repeating what I conceded already .

Let's see if we agree that speakers and room are not the end of the road for people who have an insatiable appetite for optimizing their audio systems. After all, the charter of our forum is "what is best" not "what is good enough."

 

[Kal Rubinson]

Your examples of the ear and hearing are quite old and all this work has been completed for a decade or more. We know precisely how our hearing works, its not a mystery. You may not be aware of these results, but they are known.

I wish that was so but, although we do know a lot and more each year, we certainly do not know many of the important details. I try to keep up on this for, I hope, the next edition of a textbook and there are many findings of the past 5 years that are important and relevant.

Kal

 

[RBFC]

To all of our panel of experts and industry members:

I'm placing this here, as it seems that this is one of the topics that will attract the attention of many of our experts.

As industry leaders and knowledgeable audio designers, WBF would appreciate your cooperation in an educational discussion. In your own forums, would you please describe the factors that influenced the way you have gone about developing your products? Please include both the data and the listening validation experiences as you have gone through the design process.

We believe that building an educated membership base here will produce higher quality discussions and reduce naive questioning to a more tolerable level. Just as we've attempted to create a new breed of forum atmosphere here, we would also like to continue our growth by undertaking interesting projects such as mentioned here.

So, if the science is as complete as necessary in order to design virtually "unimproveable" products, then please describe the process that led to this point. If there is much more to be learned, please describe how modifications you've made (with no solid scientific basis) have resulted in demonstrable improvements to your products. It would be valuable to see how theory is applied when designing products, and what occurs when theory is non-existent.

I believe this would be an educational bonus for our community!

Lee

 

[steve williams]

I agree. Thanks Lee

 

[Phelonious Ponk]

I once said to a friend (please forgive me if you find this insulting because he did) - show me a double-blind test for love, and I'll show you a double blind test for enjoying music.

That's a false analogy. You're not testing the music, you're testing the reproduction of a recording. It is a technical endeavor, and any valid attempt to remove expectation bias and raise objectivity is a good thing that has nothing to do with enjoying the music. I enjoy flat, limited, noisy old recordings from the late 40s and early 50s immensely.

And DBT is listening. If you truly trust your ears, you have every reason to use it often.

If the very nature of the test changes the test (quantum entanglement paradox) then the test is flawed.

Well, I don't think we need to elevate disciplined listening testing to quantum physics, but back to enjoying the music - the nature of the entire activity changes the minute we listen critically, whether it is blind or not. When we listen critically, we listen for differences, not to music. We listen to sound, for analysis, not to music, for pleasure. That does not make the listening flawed, though it certainly is, as all tests are. But if you're going to make the musical enjoyment vs critical analysis argument, the entire hobby is flawed; it's largely about critical analysis of sound and equipment. In the meantime, the only thing you can say about DBT, or even casual blind listening is that it is really the only kind of critical listening that has any chance of being objective. If you really believe you can look at your new investment of thousands, or your own speaker design, compare it to something you're already convinced is inferior, and deliver a truly objective judgement, well, I really don't know what to say.

Tim

 

[Ethan Winer]

I'm not going to wade too deeply into this, but I have to comment on what I think sets up this thread improperly from the start. From Gary's OP:

show me a double-blind test for love, and I'll show you a double blind test for enjoying music.

This is completely beside the point and has nothing to do with why we use DBT for audio. A DBT is used to tell if a difference can be discerned, and it can also help judge preference free from bias. A good example of the former is to prove or disprove claims such as whether cable elevators or replacement AC power cords change the sound. Someone can be certain they heard a change after adding a cable elevator, but a blind test (or basic audio measurement) is the best way to know for sure. An example of the latter is the loudspeaker blind tests Harmon has done, where people judge their speaker preferences without knowing the brand or cost.

To dismiss blind tests because "you can't measure love" is totally beside the point and even disingenuous IMO. Over at Gearslutz someone once posted "You can't measure a symphony" when dismissing blind tests of converter quality. So go ahead and discuss the ear, and what "science" knows or doesn't know yet. Or discuss what type of music and recordings you enjoy, which is totally different and unrelated to hearing and fidelity. But can we please avoid strawmen and other common logic errors? It will save a lot of time and effort. Here's a good reference:

Logical Fallacies

--Ethan

 

[amirm]

And DBT is listening. If you truly trust your ears, you have every reason to use it often.

Tim

I hear this often but it simply is not true Tim. DBT is a *subjective* test. Its cumulative results together with many other respondents may be objective. But the mere fact of hearing something and then voting one way or the other, still brings in a human dimension that is error prone. Take medicine. How often do we rely on the patient saying they are cured versus treating them as tools and measuring the disease in the person? Unfortunately we cannot do that in audio testing.

Further, such testing is rarely clear cut. Take this test of 24-bit, 176 KHz PCM audio against SACD: http://old.hfm-detmold.de/eti/projekte/diplomarbeiten/dsdvspcm/aes_paper_6086.pdf

100+ people took the test and didn't hear the difference to the 75% accuracy that was desired in the study. But four people did:

"The four highest scores fell into the region of “critical
probability.” This amounted to only 2.76% of all the
tests. These four tests were carried out by four separate
listeners, all of whom chose stereo music examples, and
in all four cases headphones were used—thus excluding
the influence of the listening environment to the greatest
possible extent."

A broad brush would say 98% of the trials showed that no one could tell the difference. So do we ignore the four that could? For general masses that would be wise. We know already that the larger population doesn't care and couldn't tell the difference between these two formats. But for our forum where we aspire for the best, maybe the vote of those four people should matter. And disproportionally so.

 

[Gedlee]

At the general level, I have no disagreement with what you are saying. But let's drill down. Take Steve's room. He has $30K in room acoustics, and extreme amount of money spent on speakers. I know they are not your brand but let's assume for the sake of discussion that he has achieved the pinnacle there. Should he stop there and not search for more he can do? Many of our vocal audiophiles in this forum fall in that category. They have achieved first and second order nirvana and are then searching for more as this is their hobby.

Amir - I know that you and I see things on different sides of the coin - literally. I am not rich. I seek to achieve "nirvana" on a budget that is really quite constrained. I do what needs to be done and has the most "bang for the buck" and I don't do those things that are a waste of money. Certainly there are people who are so wealthy that it makes no difference to them how much they spend. And there are certainly other people around who are more than willing to alleviate them of some of their burdonsum cash. My goal in life is to point out, and make avaialable whenever possible, how much can be done with the least expense. For those seeking to spend as much as possible, don't listen to me (just send the cash). But for those that want the best for the least, then I am their salvation.

I have never heard a system better than mine and I spent very little (considering). More times than not people tell me its the best that they have heard. Everyone agrees that what I have achieved is a miracle of cost efficiency. So yes, let's take "Steve's $30,000" plus room and system. I'm betting that I could improve it (for very little), because my science is more powerful than his checkbook.

 

[Ethan Winer]

So do we ignore the four that could?

I agree. If even one person can reliably identify a difference, then that proves the difference is audible to at least that one person. But this is no reason to minimize the importance of DBT generally. DBT is still great for dismissing (or not) the various tweaks you see me rail against here daily. If even one person can consistently identify a replacement AC power cord I'll give them $100 on the spot.

--Ethan

 

[Phelonious Ponk]

I hear this often but it simply is not true Tim. DBT is a *subjective* test. Its cumulative results together with many other respondents may be objective. But the mere fact of hearing something and then voting one way or the other, still brings in a human dimension that is error prone. Take medicine. How often do we rely on the patient saying they are cured versus treating them as tools and measuring the disease in the person? Unfortunately we cannot do that in audio testing.

Further, such testing is rarely clear cut. Take this test of 24-bit, 176 KHz PCM audio against SACD: http://old.hfm-detmold.de/eti/projekte/diplomarbeiten/dsdvspcm/aes_paper_6086.pdf

100+ people took the test and didn't hear the difference to the 75% accuracy that was desired in the study. But four people did:

"The four highest scores fell into the region of “critical
probability.” This amounted to only 2.76% of all the
tests. These four tests were carried out by four separate
listeners, all of whom chose stereo music examples, and
in all four cases headphones were used—thus excluding
the influence of the listening environment to the greatest
possible extent."

A broad brush would say 98% of the trials showed that no one could tell the difference. So do we ignore the four that could? For general masses that would be wise. We know already that the larger population doesn't care and couldn't tell the difference between these two formats. But for our forum where we aspire for the best, maybe the vote of those four people should matter. And disproportionally so.

There is nothing there I really disagree with, Amir. DBT is only conclusive if you do enough trials in a carefully controlled study, and those are pretty rare. But we'll just have to disagree on the value of informal blind listening. I think, in fact, that it may be the only effective way to listen truly critically. The psychology of bias is far too strong a force to expect any kind of objectivity from sighted listening. At least that has been my experience. I've fooled myself too many times. YMMV.

Tim

 

[amirm]

Amir - I know that you and I see things on different sides of the coin - literally.

People who follow me long term know that I don't exist on either side of the coin . As a matter of what I practice, I am far more on your side of the house though than the other. While I own fancy cables here and there I am not bothered by using ordinary ones. I have a tube amp but it is only for headphone system. All my sources are digital. You see me take the opposite side here because I feel people who are on the subjective side of these things tend to not have the engineering background so I come to their defenses if it doesn't violate my engineering sensibilities.

I have never heard a system better than mine and I spent very little (considering). More times than not people tell me its the best that they have heard. Everyone agrees that what I have achieved is a miracle of cost efficiency. So yes, let's take "Steve's $30,000" plus room and system. I'm betting that I could improve it (for very little), because my science is more powerful than his checkbook.

Steve is a gracious host and many of the members here have been to his home. I don't want to speak for him but having you there to do your thing would bring wonderful new data to our discussion!

 

[amirm]

But we'll just have to disagree on the value of informal blind listening. I think, in fact, that it may be the only effective way to listen truly critically.

The value of such testing is immense to one's self. I know. I have done blind tests on myself many times which I have documented in this forum and elsewhere. So we are in agreement there. What also needs to be said though is that such testing is boring as hell and very easy to do wrong. If this is a hobby and not search for scientific discovery, then I think people should be free to skip that step. Once there, they lose an opportunity to prove something to us but hopefully they don't live to do that .

The psychology of bias is far too strong a force to expect any kind of objectivity from sighted listening. At least that has been my experience. I've fooled myself too many times. YMMV.

Tim

I agree. But there is another technique: becoming a trained listener. Sean Olive is teaching a class in January. $300 spent on that will do one a lot better than spending the same on some tweak. There are also training material one can buy. I don't recall our trained listeners ever being wrong relative to general public in double-blind tests. And hence, we used them most of the time in our testing loop. Only when we need to be back our findings and have higher confidence did we do large scale tests.

 

[Phelonious Ponk]

The value of such testing is immense to one's self. I know. I have done blind tests on myself many times which I have documented in this forum and elsewhere. So we are in agreement there. What also needs to be said though is that such testing is boring as hell and very easy to do wrong. If this is a hobby and not search for scientific discovery, then I think people should be free to skip that step. Once there, they lose an opportunity to prove something to us but hopefully they don't live to do that .

It is boring as hell! Critical listening is boring as hell! I've reached the point where I am, at least for the time being, finished with both. My systems are not perfect, but they're excellent within their range and scale and I'm happy to just enjoy the music when I'm listening and only think about sound when I'm talking to you guys on the internet. Of course I had to add in the "for the time being." I will not be foolish enough to assume I won't hear something tomorrow that will set me off on another audio quest.

But there is another technique: becoming a trained listener. Sean Olive is teaching a class in January.

Yes, but I'm required to point out that Sean uses his highly trained listeners in blind tests. I don't think you choose techniques. I think you use both.

Tim