Why do some Objectivists fear Psychoacousitics?

[jkeny]

The difference is Harman didn't do mono tests to simplify the test analysis. They did mono tests because they worked better in comparison to stereo tests. You could say it makes the human listening easier. But we pretty well would have expected that.

Sorry, but you are just saying, in different words, that they did it in mono because it was easier to analyse. BTW, do you know if they also minimise room boundary reflections & why?

I would imagine one reason other DBT isn't done with mono testing is because of the complaints you would raise.

Huh, I don't remember being consulted by the industry? Can you tell me when this happened?

And for some purposes academic research is done in mono. It depends on what is being tested.

I believe that this answer is a yes but, no but answer - it says nothing. Again, I'll ask - if you maintain that it is generally accepted stereo is such a confounding element in test then why are listening tests done in stereo

In terms of casual testing done by hobbiests you probably would have a hard time convincing them to compare say hiresolution music to redbook by doing it in mono (I don't know if that would work better). But hobbiests in high end audio are often difficult to convince about a good many absurd beliefs they hold.

Well demands are made for volume matching, & recently for having someone oversee the test as gaming is suspected so what's the problem with suggesting mono listening when you have a scientific research proving it is a more effective at discriminating differences? `

It would be good to have others replicate Harman's work to see if it holds up or not. But Harman came to these ideas through plenty of work on their part. It wasn't like they set out with predetermined ideas of what was best. It is what their work and testing taught them. Further while they talk about it some it hasn't been any big part of their marketing. As in ad copy. They market their speakers much like anyone else. It would appear they believe their testing has let them build better speakers, and given everything in total they believe better speakers at a given price point will earn them more marketshare. So they are marketing the speakers not the testing benefits for design purposes

I didn't mention anything about Harmon's marketing - I wondered how reliable a single piece of unreplicated research is - the scientific principle would put it in the category of interesting ideas but not "proven"

Using simple tests, learning how things work, what is important piece by piece is an old time honored successful idea.

Of course it is but the parameters & limitations of such testing is also highlighted in any reasonable conclusions to such testing. This is the missing perspective I see.

Jumping whole hog into something and insisting you have to get it all in one go in totality or not at all has a history of slow or no progress with lots of confusion about what is important along with confusion about what works and what doesn't. When an area of life transitions from the old way to the newer way based upon investigation it is a repetitive story to see those decry the testing because the results disagree with previously held though foggy ideas about something. Though the early period of progress can sometimes be two steps forward and one back, even that is tremendously more rapid and effective than what has gone before.

Yes & I'm saying that we have gone about as far as we can go with simplistic tests which ignore psychoacoustics (admittedly a difficult area to grapple with) but instead of embracing this suggestion all I see from you is push back. I would have thought that true objectivists would be looking for ways of advancing closer to measurements/tests which better represent our auditory perceptions instead of trying to defend the existing test & measurements.

 

[amirm]

Here is some of the data with regards to mono vs stereo testing.

Comparison of Loudspeaker-Room
Equalization Preference for Multichannel,
Stereo, and Mono Reproductions: Are
Listeners More Discriminating in Mono?
Sean E. Olive, Sean M. Hess, and Allan Devantier

Harman International, Northridge, CA, 91329, USA

The setup was a test of room equalization systems. First phase determined that all three EQ strategies garnered positive preference among listeners. Performing no eq was the reference/control.

The second phase then tested the number of channels in each scenario including the no eq/control:

Notice the massive drop in the scores of no-eq configuration as the number of channels drops to mono. As the number of channels increases, the improvement over doing nothing becomes smaller and smaller. Put inversely, whatever ills the speaker had was most apparent in mono. As you increase the channels, the listeners tend to not pick up on their flaws as much.

Here are the conclusions from the paper itself:

The main conclusions to be drawn from this listening
experiment are the following:

1. All three equalization methods were preferred
over the unequalized version of the loudspeaker.

2. There was no significant preference among the
three loudspeaker equalization methods: direct,
predicted-in room, and in situ.

3. There was a significant interaction between the
different loudspeaker equalizations and the
number of playback channels. The effect was
largely isolated to the unequalized loudspeaker;
its preference rating decreased monotonically as
the playback channels were reduced from five
(surround), to two (stereo), to one (mono).

 

[jkeny]

There is a difference between accurately reproducing a recorded signal and the methods used by the engineers to mix and master that recording to sound pleasing or to some realistic. There can be no other stance to take except that the duty of the two channel system is to reproduce the signals as exactly as possible. To make two channel sound good is the job of the mix and master engineer, not the electronics, and technically not the speaker or mic, those things all need to be as perfect as possible, period. The art, the psychoacoustics is not the responsibility of the two channel system, as it does what it was supposed to do, that is allow your ear/brain interface to believe there are sources of sound between the speakers. Just sayin.

Well I was talking about "accuracy" to the recorded signal on disc, PC, vinyl, etc. , not necessarily to the original audio event.

But this accuracy is still not completely achievable so we choose compromises. The compromises/inaccuracies should be chosen which are least damaging to our auditory perception. This is where the psychoacoustic considerations come into play.

My contention is that our replay systems are still violating some psychoacoustic principles & these are yet to be discovered. When they are we will have a more realistic illusion. We will then know the true limitations of a stereo reproduction system.

 

[Don Hills]

Here is some of the data with regards to mono vs stereo testing.

...
3. There was a significant interaction between the
different loudspeaker equalizations and the
number of playback channels. The effect was
largely isolated to the unequalized loudspeaker;
its preference rating decreased monotonically as
the playback channels were reduced from five
(surround), to two (stereo), to one (mono).

I read that as saying that changes to the sound (different EQs, and presumably by extension different loudspeakers) are more obvious when listening in mono compared to stereo or 5.1. Do I understand correctly?

 

[Robh3606]

I read that as saying that changes to the sound (different EQs, and presumably by extension different loudspeakers) are more obvious when listening in mono compared to stereo or 5.1. Do I understand correctly?

That's the way I read it and it makes sense. Similar to multiple subs below the transition frequency

Rob

 

[esldude]

I read that as saying that changes to the sound (different EQs, and presumably by extension different loudspeakers) are more obvious when listening in mono compared to stereo or 5.1. Do I understand correctly?

Now for jkeny's interpretation.

Extra channels are conveying more information than overly simplified mono or even stereo vs surround. Therefore the listeners were more able to perceive what the music was supposed to sound like due to the psychoacoustics of the situation. As more channels means the speakers responses are not as relevant to the listener. Meaning that more channels and more information clearly are not understood with such simplified testing as it stands now. More research needs to be done, and not this kind of flawed approach.


Sorry, jkeny, couldn't resist.

 

[jkeny]

Thanks Amir
This is a conclusion from that paper

“The third conclusion that listeners are more discriminating of spectral defects in mono versus stereo and multichannel playback modes is both thought-provoking and reassuring. It adds confirmation and validation for continuing subjective evaluation of loudspeakers in mono when the most sensitive and critical sound quality assessments are required. Not only are mono listening tests more revealing of loudspeaker-room artefacts, they are eminently more practical, efficient and cost-effective to set up and administer.”

My question is, are "spectral defects" the only defects of importance in loudspeakers?

Don't get me wrong, I applaud Harmon for their tests but that doesn't mean that by asking questions about their conclusions that I'm dismissing their results or efforts - I'm just interested in the limitations of their results

 

[jkeny]

Now for jkeny's interpretation.

Extra channels are conveying more information than overly simplified mono or even stereo vs surround. Therefore the listeners were more able to perceive what the music was supposed to sound like due to the psychoacoustics of the situation. As more channels means the speakers responses are not as relevant to the listener. Meaning that more channels and more information clearly are not understood with such simplified testing as it stands now. More research needs to be done, and not this kind of flawed approach.


Sorry, jkeny, couldn't resist.

No, what I'm saying is that it's the equivalent of judging 3D TVs with on eye closed
Spectral differences will be picked up but maybe, just maybe that isn't all there is to our stereoscopic vision
Of course for people with one eye that is fine

 

[Robh3606]

No, what I'm saying is that it's the equivalent of judging 3D TVs with on eye closed

Hello Jkeny

Yeah but what I think some miss is that the sound sources in the real world are mono. That wasn't two birds chirping 8 feet apart it was one. Same for the twig snap, dog bark, person talking, and so on. It doesn't seem strange when you think of it that way. We are too fixated on our mode of reproduction. Stereo is not a real thing for sources just for how we hear them in the sense we have a pair of ears.

Rob

 

[FrantzM]

No, what I'm saying is that it's the equivalent of judging 3D TVs with on eye closed
Spectral differences will be picked up but maybe, just maybe that isn't all there is to our stereoscopic vision
Of course for people with one eye that is fine

It is conjecture John. The fact remain that when the knowledge is removed our opinions tend to change. That is a fact not an opinion, not a conjecture. If we are about opinions, there is no way I can discuss your opinions only saying yours are yours and mine is ...

So far it has been a lot of conjecture on your part. The facts (if any)have been sparse. That we don't know it all is a truism there will always be something new to consider, it doesn't invalidate that our knowledge of a given product is the greatest bias and our response to our biases goes below our level of consciousness . So for once let us leave the truisms aside.

One more thing: We are very clear about the limitations of 2-channel or MCH for that matter, reproductions. They are lacking and we know it the second we compare live to any stereo system. Eye shut , wide-open or wide-shut

 

[jkeny]

Hello Jkeny

Yeah but what I think some miss is that the sound sources in the real world are mono. That wasn't two birds chirping 8 feet apart it was one. Same for the twig snap, dog bark, person talking, and so on. It doesn't seem strange when you think of it that way. We are too fixated on our mode of reproduction. Stereo is not a real thing for sources just for how we hear them in the sense we have a pair of ears.

Rob

Yes, Rob, but the whole stereo reproduction chain's job is attempting to recreate an illusion of a real audio event (or an interpretation of that event) using more than one microphone & using various studio techniques to portray a sound stage - if you ignore all this by ignoring half of that signal on replay, I don't believe you are evaluating it accurately & any preferences will, as a result, be skewed. Harmon/Olive/Toole say they they won't be skewed but I'm not yet convinced of this. Having confirmation from other research would help but it seems there is none

 

[jkeny]

It is conjecture John. The fact remain that when the knowledge is removed our opinions tend to change. That is a fact not an opinion, not a conjecture. If we are about opinions, there is no way I can discuss your opinions only saying yours are yours and mine is ...

So far it has been a lot of conjecture on your part. The facts (if any)have been sparse. That we don't know it all is a truism there will always be something new to consider, it doesn't invalidate that our knowledge of a given product is the greatest bias and our response to our biases goes below our level of consciousness . So for once let us leave the truisms aside.

One more thing: We are very clear about the limitations of 2-channel or MCH for that matter, reproductions. They are lacking and we know it the second we compare live to any stereo system. Eye shut , wide-open or wide-shut

Frantz, you quoted my post but then ignored everything in it in your answer.

If you find my analogy wrong - it is the equivalent of judging 3D TV with one eye closed - then please tell me why?
Stating that it is all conjecture on my part really doesn't get us anywhere.

Talking about blind testing - again I find that always the focus is on what bias is REMOVED rather than what extra influences(Biases) are INTRODUCED by the blind test. I'm constantly trying to direct people's attention to this but there's a great unwillingness for them to do so. The presence (or not) of these EXTRA influences could easily be demonstrated by the use of controls for false negatives in the test but again there is a great unwillingness to do so.

Until such time as I see this I will always have my suspicions that blind tests skew results towards not hearing differences which actually exist. The Harmon graph certainly shows that in blind testing the preference between different speakers was very slight - does this correlate with the actual measured differences between those speakers? In other words given measurements that fully characterised each speakers output, would it be predicted that there would be very little audible differences between them - they would all be similar in preference?

I've also given my reasons why A/B testing ignores the more important (for our enjoyment) psychoacoustic area of our hearing & I believe Harmon are making the same mistake.

These psychoacoustics are not conjecture (In so far as any scientific research is) - they are the best model we have so far for how auditory perception works. Auditory Scene Analysis is not conjecture - it has an equivalent in Visual scene analysis (a fairly well studied & understood area, I believe) & has similar operational aspects.

You may believe that we know all there is to know about the limitations of 2 channel & MCH but I don't think that is correct.

 

[jkeny]

I can tell you that in my listening, the things that make the most difference, are how realistic the sound is & this, I believe is because the system is doing something which better appeals to our psychoacoustic model of reality.

- when I recently heard a Lampizator 7 playing DSD, what immediately struck me was how much each element in the sound stage remained solid & real - so in the presence of loud passages, the quiet elements were still undisturbed & realistically presented. As a result the sound stage was solid, real & not modulating - better sense of believability.

- in all my testing/modifications/development, it's this that is the main aspect of what I'm looking for as I believe we have mostly moved past spectral differences (in electronic devices) & should be looking at areas of realism (psychoacoustics) which are perhaps considered more subtle (because they are not as easy to test/analyse) but are actually more important for our long term listening & enjoyment (isn't this what the hobby is about?).

I would be surprised if other developers weren't following this path also

 

[jkeny]

In following up some links, I came across this quote from Toole's paper

Using the highly reliable subjective ratings from an earlier study, loudspeaker measurements have been examined for systematic relationships to listener preferences. The resuls has been a logical and orderly organization of measurements that can be used to anticipate listener opinion. With the restriction to listeners with near-normal hearing and loudspeakers of the conventional forward-facing configuration, the data offer convincing proof that a reliable ranking of loudspeaker sound quality can be achieved with specific combinations of high-resolution free-field amplitude-response data. Using such data obtained at several orientations it is possible to estimate loudspeaker performance in the listening room. Listening-room and sound-power measurements alone appear to be susceptible to error in that while truly poor loudspeakers can generally be identified, excellence may not be recognized. High-quality stereo reproduction is compatible with those loudspeakers yielding high sound quality; however, there appears to be an inherent trade-off between the illusions of specific image localization and the sense of spatial involvement.

I'm not sure how to read the bolded statement - anyone help in understanding this?

 

[FrantzM]

The notion of "realistic" is subjective, thus the plethora of products on the market .. All of them claiming to be more "realistic" than the others. There is no way anyone can tell you that Blind Testings are perfect. They can't be but in this world of imperfection, some "things" are more imperfect than others. On that, knowledge of the DUIT has been proven to be strong. Not conjecture has been proven. You can speculate what other bias not knowing introduces... and that will remain that a speculation, a conjecture , not a fact.

The enjoyment argument is weak IMO. 99.99% D) of the reproduced Music Listening population enjoy music on things that no one would dare label as realistic reproducers of music (iPod with the Apple ear-buds ... for example) So those manufacturers could be doing something right if we use those metrics ... 128 kbps mp3 would have won that war ...

There is however something that strikes a note in my music lover mind. If you are just talking about altering the signal in such a way that is become more pleasing to some people, then that becomes a worthwhile pursuit... Interestingly your strongest ally in that endeavor is again ... blind testing. The subjects must not know of the alteration else it becomes a matter of basic, well known psychology ... suggestions and you will admit that these are the strongest biases known to men ...

Oh and I don;t believe in complete knowledge no one ever implied such ... Argumentation fallacy if we were ever looking for one.

 

[andromedaaudio]

Measurements only tell so much , a flat FR can be achieved with a LS costing 500 $ .
What that does not tell you is ; correct phase , does the membrane ring , has the membrane a natural sound , how much does the housing interfere with the signal and so forth...
I heard the revels salons on a couple occasions and i found them not very special sounding , the membrane material they used sounded artificial not natural , plus they sounded "bass light " to my ears
A flat loudspeaker to , for example 25 hz will sound basslight especially if multiple smaller bassunits are used instead of a big one (preferably playing a couple of db s above flat )
With a high end ls costing 20 K a flat FR from lets say 350 hz and up should be mandatory , a minimum of +- 2,5 db 1/6 octave smoothed , if not the LS does indeed not sound tonally correct.

 

[amirm]

TMy question is, are "spectral defects" the only defects of importance in loudspeakers?

They constitute colorations which are the easiest to hear. And clearly don't make for good speaker that is reproducing the source.

Don't get me wrong, I applaud Harmon for their tests but that doesn't mean that by asking questions about their conclusions that I'm dismissing their results or efforts - I'm just interested in the limitations of their results

No worries.

 

[amirm]

If you find my analogy wrong - it is the equivalent of judging 3D TV with one eye closed - then please tell me why?

If the flesh tones are green, 3-D is not going to save it. But might throw you off at first with the amazement of seeing 3-D images.

An actual video analogy might help. If you take a video system to broadcast professional, they usually opt to turn off the color and evaluate the luma (black and white) image first. For this reason, all professional monitors have switch that does exactly that: turns off color. That way the prettiness of the color is removes and with that we can more clearly see the artifacts in the luminance channel. Since the eye is more sensitive to black and white images than color, that is an important aspect of evaluation.

Likewise here, by listening to mono instead of stereo, and focusing on colorations, we have a set up that is more revealing of the most objectionable flaws of speakers.

When I sat through Harman test and listened to the same clip over and over again with different speakers, you immediately got a sense of how true the above is. Voices for example sounded drastically different from speaker to speaker. It is remarkable that this day and age there is so much variation in reproducing that all important sound. Surely we want our vocals to sound true. Any distortion if present outside of coloration was not even registering in massive tonal variations introduced by the speaker.

 

[jkeny]

The notion of "realistic" is subjective, thus the plethora of products on the market .. All of them claiming to be more "realistic" than the others.

No, Franz, this is where we fundamentally disagree - Realism comes from our auditroy perception which follows the same rules for everybody - it's why we can all tell real world audio from reproduced audio, as you stated before. I agree with you that getting the illusion closer to realism requires a lot of complexity & this, I suggest is why there are so many products on the market.

There is no way anyone can tell you that Blind Testings are perfect. They can't be but in this world of imperfection, some "things" are more imperfect than others. On that, knowledge of the DUIT has been proven to be strong. Not conjecture has been proven. You can speculate what other bias not knowing introduces... and that will remain that a speculation, a conjecture , not a fact.

But, it has always been part of blind testing that internal controls should be used specifically so that there could be no conjecture over the test itself as to it skewing results towards a null. So don't blame me for my doubts - blame those who organise blind tests. No amount of claims that my views are "conjecture" will remove the doubt that all people should have about these tests (tests that don't have internal controls)

The enjoyment argument is weak IMO. 99.99% D) of the reproduced Music Listening population enjoy music on things that no one would dare label as realistic reproducers of music (iPod with the Apple ear-buds ... for example) So those manufacturers could be doing something right if we use those metrics ... 128 kbps mp3 would have won that war ...

Well, if you deny that better reproduction of audio increases our enjoyment/involvement/understanding of the piece then, again, we fundamentally disagree

There is however something that strikes a note in my music lover mind. If you are just talking about altering the signal in such a way that is become more pleasing to some people, then that becomes a worthwhile pursuit... Interestingly your strongest ally in that endeavor is again ... blind testing. The subjects must not know of the alteration else it becomes a matter of basic, well known psychology ... suggestions and you will admit that these are the strongest biases known to men ...

No, these are old arguments about "preference" - I'm asking the question "Seeing as we don't know the rules involved in auditory processing, how can we say that our systems have reached their limitation in how well they match to these rules?" The other thing I'm saying is "that A/B testing probably ignores a large part of what is important - the psychoacoustics of what we are listening to. A sound stage, for instance, doesn't suddenly appear to us when we switch to listening to a device - it takes time to be built up in our auditory perception & this relies on many aspects of the sound, ambience, etc. Therefore quick A/B switching, which adheres to the audio echoic memory limit of 4 secs is unsuitable for examining differences in this aspect of the sound. There are many more areas of psychoacoustics that A/B testing is not suitable to examining".

Oh and I don;t believe in complete knowledge no one ever implied such ... Argumentation fallacy if we were ever looking for one.

Sorry, if I overstated your statement "We are very clear about the limitations of 2-channel or MCH for that matter, reproductions." into "You may believe that we know all there is to know about the limitations of 2 channel & MCH but I don't think that is correct." so between these two statements, I'm saying that there is some areas of the limitations of 2 channel reproduction that are unknown - unknowns are in the area of how well our reproduction system adhere to psychoacoustic principles.

 

[jkeny]

They constitute colorations which are the easiest to hear. And clearly don't make for good speaker that is reproducing the source.

Indeed & that is important but is it the MOST or ONLY important aspect - that is my question! It obviously isn't & I included another quote (in another post) from Toole that seems to address this but I don't understand what it means