Audio Asylum Thread Printer
Get a view of an entire thread on one page
|For Sale Ads|
Here's a question on something I really don't know a thing about:
In this era of hi-rez sound sources like SACD and DVD-A, what's the significance/impact of extended frequency range for/on the perception of speaker sound?
As we know, human hearing is more or limited to below 22 kHz, and so most speaker manufacturers, rationally enough, have been happy to settle with the 20-20,000 Hz range. But a lot of today's premier models are capable of notably good reproduction throughout ranges up to 40 kHz and even 100 kHz, using the newer generation "supertweeters" and other innovative engineering solutions (makers such as Vivid, Gamut, Marten, and Avalon come to my mind). These speakers seem to garner nothing but highest praise in listening tests, especially for their exquisite highs and the impact of these highs on the overall experience of the reproduced sound. Is this reaction simply nothing but a reflection of the great overall engineering job done (flat freq. response way up), or does the hyperextended upper range indeed somehow specifically matter for the way we perceive the sound, even if we can no longer hear it?
That's of course much in the context of SACD and DVD-A recordings whose response exceeds the range of our hearing.
The same question could probably be asked about preamps, too. Why do their range need to extend way beyond the limits of our hearing?
Sorry if this speaks of nothing but blind ignorance on my part.
If I remember correctly that study contains, according to Kal Rubinson, who is degreed neuroscientist, methodological flaws.
I've read that paper, can't remember in detail what's in it, did they test the supersonic signal on its own and measured the response?
As for the highest praise in listening tests, keep in mind that probably none of these were done under controlled conditions, so the results are questionable.
Oohashi published two papers on this issue, the first at a 1991 AES convention (no peer review), the second in J. Neurophysiology 2000 (peer-reviewed). In both papers the HF content alone is presented to the subjects and response is measured.
In this respect, the AES paper describes two experiments, one with genuine Gamelan music, the 2nd with simulated Gamelan music, derived from white noise.
In the 1st case, 6 out of 16 subjects responded to the HF signal, in the 2nd case only 2 out of 16. The paper does'nt provide any statistical analysis whatsoever.
The 2000 paper presents the HF results in statistical terms only.
Anyway, I think that such studies are, as interesting as they may be in themselves, of little value to the music lover since hi-rez audio formats appear to have no commercial success.
You are right about the last point...
But the 2000 paper (or am I totally misremembering it from yesterday?) beamed to the subjects the same material thrice: (1) audible range only (below 20 kHz); (2) HF content only (above 20 kHz); and (3) "full range" (both). The response appeared only in case (3), but only in the form of a subliminal physiological reaction and subjective listening preference as gathered from the questionnaire responses. In no case did the subjects report "hearing" any HF content.
I am not convinced that the HF content produced no IM distortion effects in the audible range, and hence, it does not seem to me that ultrasonic hearing is established by Oohashi et al. Kaoru and Shogu used more than two drivers, I think. Also, Kaoru and Shogu showed measurements to determine whether there was any IM distortion produced in the audible range.
"Opposition brings concord. Out of discord comes the fairest harmony."
------Heraclitus of Ephesis (fl. 504-500 BC), trans. Wheelwright.
Just curious, do you know how "detecting" the LF/HF contents in the audio clips was confirmed in their study? Was this detection about "hearing" the signal (the usual way, through ears) or did they use some other form of physiological monitoring?
I couldn't find this paper online except as a $20 download, so I
only read the abstract.
In either case, the possible presence of IM distortion might mean many things or nothing really, but it's hard to say without knowing the test setup and procedure.
...the way we perceive sound is very complex.
When a small change is made at one fequency, it affects our perception of other frequencies, sometimes in the opposite end of the frequency spectrum.
You can experiment with this if you have a speaker with tweeter or supertweeter controls.
This seems to occur even if the added information is outside the normal audible hearing range.
When people add supertweeters, they usually report an increase in sound quality overall including the bass frequencies.
Audiophiles report almost EVERY new component they spend their money on increases their sound quality enjoyment. That doesn't mean there's any audible change at all. It just means audiophiles and other humans tend to defend their purchase decisions.
Your objection is about an entirely different issue (psychology of consumer behavior).
This was about human responses to a certain kind of acoustic stimulation. The study was controlled, and clearly shows that in the cases in which there were responses in deepseated (subliminal and autonomous) brain functions (as shown by changes in blood flow and electrical activity), the listeners also subjectively felt something different. And what this "difference" is about we don't really know, but the listeners articulated it as a feeling of personal preference (greater enjoyment, summing up from the questionnaire replies). I thought it was interesting.
What's the troubling part of this? Just curious about your strong reaction.
No one ever said all humans have hearing that goes to 20kHZ. and then NOTHING above is audible no matter how loud.
The unusual "music" chosen was something that few, if any, audiophiles coming here have heard of, or would listen to.
It was chosen for unusually loud high frequency content and may have no correlation with conventional Western music, making the experiment moot for those who don't listen to this particular "music".
You've also fallen into the trap of assuming that when people say they like A better than B, that means they can really hear a difference.
That "logic" skips one important step in the "assumption ladder"
In high end audio, people will say they like A or B between 50% and 75% of the time when A and B are really the same component playing at the same SPL, merely switched off and on to suggest to the listener there are two different components being compared. So what do those "A or B" preferences prove?
Just a quick point or two:
That "music" probably wasn't as "unusual" and weird for the test subjects as it might be for you. (They were Japanese.) You shouldn't get arrogant about your own musical preferences; they are just that, nothing more, even though culturally largely (pre)determined.
And as for "audiophiles," I think to all accounts in Japan you'll be likely to find more of them, and more serious ones, than you do in the U.S. They'd probably think very differently about the suitability of Southeast Asian classical music for audiophile ears.
Moreover, as a footnote one could add that gamelan music has been incorporated into "conventional" Western classical music as well, and wouldn't represent that bizarre an encounter to any friends of Messiaen at least.
"You shouldn't get arrogant about your own musical preferences; they are just that, nothing more, even though culturally largely (pre)determined."
The music for the test was chosen for testing audibility of high frequencies because it had an UNUSUAL amount of high frequency energy.
That could bias the test.
Obviously with louder/more high frequencies, it's more likely the high frequencies would be audible.
As a result, the test results may not correlate to doing the same test using ANY other types of music without an unusual amount of high frequency energy.
My statement was about the music selections used and their potential correlation to any other music without so much high frequency content.
The test music is also unlikely to be in the collections of people visiting this web site, or is likely to be a very small percentage of the music collection.
YOUR decision to call me "arrogant" could only be based on your misunderstanding of the science involved. To choose music with an unusual amount of high frequency energy for a test of high frequency audibility cleary biases a test.
For whatever it's worth, there's quite a bit of extreme-HF energy involved in just about any standard orchestra's playing conventional Western music:
If the point is to measure HF content, you find material that has that HF content (and is not too distracting to the test subjects), simple as that.
Have no intention to insult anyone here, and will simply trust that neither do you, behind the letter of your posts.
... unless you sit four feet directly in front of the trumpet player (you'd get hearing damage if you did) ... so the data have no correlation with what a real person would hear at his seat in an auditorium listening to a real orchestra ... even if he sat in the front row.
Richard BassNut Greene
Subjective Audiophile 2007
I don't know who you are arguing with.
Just to repeat: no one here thinks this line of questioning is directly applicabile to audio engineering in some unproblematic way that has immediate consequences for consumer designs. No one thinks we should all run to the store to buy a supertweeter all of a sudden. And none of this has anything to do with hearing, either, for the n'th time.
The issues are hypothetical and that's where their interest lies. David A. has summarized this intelligently below. For me (and I'm assuming you are interested in my perspective since you seem to believe you are somehow arguing against my take on this), it is the what-if type of questioning that gives the fascinating angle on all this. If moderate LF content enhances listening pleasure, and extreme LF content can induce neuropsychological effects on "listeners" (inaudible subsonic vibration resulting in anxiety reactions in test subjects, etc.), could we then ask whehterhypersonic HF content, too, might have some comparable but maybe qualitatively different effect on the experience of sound perception.
What sort of an effect might that be? What prompted me to think about all this in the first place (in the context of hi-rez audio) was the uniformly positive response that some new speaker designs utilizing "supertweeters" have been met with in listening tests; a response moreover that seems to steer away from the banal "liquid midrange" and "smooth highs" discourse in describing the listener reactions. The auditing experience is usually summarized with the rather simple bottom line as being "very, very nice" in some hard-to-pin-down kind of a way (the point being about noteworty presence of subjective listening pleasure). What if in some way it's about the HF content that these models can reproduce?
Whether this hypothetical effect has anything to do with IM distortion is another matter. If it does, what would happen if through design solutions this distortion could be eliminated? Would the effect then be equally or even more pronounced? Or is it precisely IM distortion that causes this response in the subjects in the first place - in which case we could maybe conclude that here we have yet another possibility for euphonic solutions in sound reproduction devices? Or is the whole IM distorion issue something that exists apart of the engineering matrix, as some general property of soundwave behavior in the atmosphere? (In which case it would be interesting to consider why we have the perceptive capability for it somehow built into the design of our anthropological apparatus.)
The more directly consequential questions for someone thinking of a speaker upgrade lie obviously elsewhere, most immediately in things like the accuracy of frequency response within the audible range, lack of coloration & distortion, and good interaction with the environment in which the system is set up. Without these you will have no satisfying sonic experiences whatsoever.
It's important however to remain open to questions and have some fun in the process even if the issues seem purely speculative at present. Who knows, maybe tomorrow they are a feature you can choose for your system setup, too. Obstinacy combined with a single-issue mindset is never a good thing to have, and we all know we have enough of ideology and dogmatism in this field already (for the record, I'm not claiming these are what you stand for). Why not be open-minded and exploratory? The research problem in the study was intriguing, pregnant with implications, and it certainly merits further consideration.
But enough of this already.
I have no idea why you are debating me over mkuller's post saying that supertweeters get good reviews from their owners and also claiming a Japanese study could explain why.
I responded that most people defend their purchase decisions -- it's human nature -- so that doesn't mean much.
I also added the study was biased and NEVER even tried to prove the listeners could hear a difference between the music with, and music without, frequencies over 22kHz.
The study leaders were too eager to accept subjective "preferences" (non-science) without testing to see if the listeners could hear any difference under blind conditions (real science).
The study has very little connection with the audibility of adding supertweeters.
I rest my case.
Assuming that's true, does anyone know what this effect is based on, since we really can't hear that part of the bandwidth?
Those make for a good read. I think this is an interesting subject that has obvious and important implications for speaker design. The evidence, at least in these well-referenced and to all accounts solid research papers, seems indeed to indicate that inclusion of above-22kHz high frequency information in the audible bandwith, even when imperceptible as such, produces neurophysiological responses in the brain and a psychological effect in the listener that amounts to a subtle sensation of pleasure and subjective satisfaction - in other words, perhaps the listener happiness that we all strive for through all that gizmo swapping and system tweaking upstream?
That would also explain why the sound of the speaker models I mentioned is found extraordinarily pleasing in prolonged listening tests, whereas for example some studio monitors (which usually stay within the sub-22kHz range) appear "detached" and subjectively "uninvolving" even when their respective performance capabilities would otherwise be equally or more impressive in key respects and based on superior engineering solutions. (I know, there are just too many other variables involved, too.)
Thanks for responding.
Participants never proved they could hear a difference.
They merely voted on which version of the Bali "music" they preferred.
That proves absolutely nothing.
Random guessing could easily lead to people preferring either A or B
BUT THAT IS NOT PROOF THEY CAN HEAR ANY DIFFERENCE AT ALL.
People will choose A or B because they are trying to please the experiment leader -- rather than saying "I don't know", or "I can't hear a difference". True of high end audio too!
The psychological assessment was just one (and subordinate, sort of control) part of it. What correlated with it (and took priority over it in the evaluation) was two other things: an EEG analysis and a PET measurement in each case. The subjects didn't need to hear any difference and it was even assumed they wouldn't; their bodies registered the variations subliminally, and the structured interview questions confirmed that they indeed had also sensed or perceived (not heard, since this was all ultrasonics) something.
The study seems important in the field - even at a superficial look around you can see that it's frequently cited in serious studies and by serious people whom I think you, too, would recognize as an authority.
I think you just have a big bass bias!
If I remember correctly, some time ago I have done a search in the Science Citation Index and could not find any paper that cites the Oohashi papers, other than Oohashi himself. I can do that search again next week just to check.
I need to do a better check: for now I just ran a quick Google search and found citations by people such as David Griesinger (of Lexicon), James Boyk (director of Cal Tech music lab and lecturer in its dept. of electrical engineering), and a bunch of faculty at psychology, music, and various tech depts. around the States(Berkeley etc.). Plus a co-authored piece in an audio engineering journal in Germany (which I can't recall now). These were all references that seemed to take the Oohashi et al. paper seriously. And then of course the references in Stereophile and other audio fora.
From my quick search I was not able to find any refutations of either the hypotheses or the research procedure, though that may be because the research questions simply haven't been adequately put to test yet. (Meaning it's just too hard to evaluate, apart from the test setup and methodology, but even they seemed fine at a quick glance.) What seemed to be a novelty in the Oohashi 2000 paper is that the audio clips were longer, as were the pauses in between them; the assumption being that certain physiological responses only become manifest rather slowly (compared to the usual few seconds done in audio tests) and have a correspondingly slower settling time afterwards (have a hang time).
I would think J of Neurophysiology must be quite a respectable publication in the field (of which I know nothing to be sure). There was moreover another paper by Oohashi in 2006 in Elsevier's Brain Research journal, 16 February 2006, in which the conclusions are that this HF effect with a bearing on listening preferences and manifest in brain function responses may not have anything to do with "the conventional air-conducting auditory system," since this effect was manifest "only when the listener's entire body surface was exposed to High Frequency Components, but not when HFC was presented exclusively to the air-conducting auditory system." (They used a loudspeakers vs. headphones type of a setup.)
Oohashi himself has a nice career behind him (from Wikipedia - ):
"He held positions such as Instructor at Tsukuba University, Professor at the National Institute of Multimedia Education, Professor at Chiba Institute of Technology, and General manager at the Department of KANSEI Brain Science, ATR Human Information Processing Research Laboratories. He is the President of the Yamashiro Institute of Science and Culture and the Director and chief researcher of the Foundation for the Advancement of International Science."
I just find this quite intriguing as a hypothetical issue. Not sure what the repercussions would be for audio design; there are just too many variables involved, even if Oohashi & co. are right. (The research reported in the J Neurophysiology issue was based on double blind testing and the results were statistically significant.)
More stuff to read, that is. What's your take on all this?
"I need to do a better check: for now I just ran a quick Google search and found citations by people such as David Griesinger (of Lexicon), James Boyk (director of Cal Tech music lab and lecturer in its dept. of electrical engineering), and a bunch of faculty at psychology, music, and various tech depts. around the States(Berkeley etc.). Plus a co-authored piece in an audio engineering journal in Germany (which I can't recall now). These were all references that seemed to take the Oohashi et al. paper seriously. And then of course the references in Stereophile and other audio fora."
These are not the type of references that lend credibility to a study. These references are merely using the results/conclusions of the study to support or interpret their own interests and phenomena. What is singularly lacking is any independent duplication and confirmation of Oohashi's work. That does not mean it is wrong but, rather, that it is unconfirmed and no other competent neurophysiologists seem excited enough to do so.
"I would think J of Neurophysiology must be quite a respectable publication in the field (of which I know nothing to be sure). There was moreover another paper by Oohashi in 2006 in Elsevier's Brain Research journal, 16 February 2006, in which the conclusions are that this HF effect with a bearing on listening preferences and manifest in brain function responses may not have anything to do with "the conventional air-conducting auditory system," since this effect was manifest "only when the listener's entire body surface was exposed to High Frequency Components, but not when HFC was presented exclusively to the air-conducting auditory system." (They used a loudspeakers vs. headphones type of a setup.)"
This seems to be consistent with the strange results of Oohashi's localization of brain responses to regions which are, to a great degree, not generally associated with audition.
Instead of ripping apart a series of unduplicated reports, I'd rather regard them as intriguing and await independent confirmation.
Sure, except in your point that these authors are "merely using the results/conclusions of the study to support or interpret their own interests and phenomena." I think that would be hard to prove, and it's not exactly the kind of activity these people (Griesinger, Boyk & co.) are known for in the audio engineering circles (unless you are referring to those folks at Stereophile). We can safely say that at least those named are more scientifically oriented in their methods and approach than the great majority of those earning a living from audio industry out there. Moreover none of those I mentioned, as I think I stated, framed the findings of Oohashi et al. in any way; they merely observed the existence of this research and its results, calling attention to the broader issues it too addresses. Sort of tossing it in the air and asking: What do we make of this?
So I'd say that's a bit biased as a statement from you.
Nope. None of these people, although respected and respectable in their own fields they may be, are neurophysiologists. Since they cannot replicate these or similar experiments, their use of the results does not constitute confirmation of the findings. When one looks at citations, those of reputable scientists in the same field are significantly different from those from outside the field. So, no disrespect for these people was implied.
OTOH, I have yet to stumble on any references to Oohashi's work in the neurophysiological literature. The latter would be of interest to me.
Not to be splitting hairs, but your initial claim was that they are "merely using the results/conclusions of the study to support or interpret their own interests or phenomena." That's a positive statement: it's basically claiming that they *are exploiting* the study for their own purposes. That's a very different claim than your current negative statement that "their use of the results does not constitute confirmation of the findings," which proposes that they *are not validating* the study.
The thrust of these two statements is very different in each case. I wouldn't disagree with the latter but I'm frankly puzzled by the former.
Without taking a stance on the merits of the paper (apart from my fascination with its research problematic), I haven't found anyone either criticizing its procedure or repudiating its results (apart from these very pages, that is). Since in accordance with all basic principles of scientific method the paper is very open in recording the details of its test setup and procedure, in principle anyone can also verify its basic correctness and assess its fundamental flaws. Instead of vague hints about lack of response in one academic corner (but not in some others) constituting a de facto refutation, why not break the silence and tell what's wrong with it, anyone? The floor is open and the audience is waiting and I am already busy making notes.
You're interested in seeing a credible confirmation, I'm interested in seeing a credible refutation; at some point we must compare notes.
I think you are splitting hairs. The reference and use of the conclusions of a scientific publication is fair game for anyone. For these audio technologists to do so is perfectly reasonable and I did not imply any misuse on their part.
As for the second issue, the lack of any scientific interest in this subject is, especially in the face of Oohashi's publications, to me, telling. It is suprising that it did not stimulate any interest in the field of auditory or sensory neurophysiology.
I have not read the paper carefully in quite a while but I recall that my procedural objections (and surprise at some of the results) do not invalidate the paper. They only prevented me from accepting it fully until there is some independent confirmation.
"You're interested in seeing a credible confirmation, I'm interested in seeing a credible refutation;" Frankly, I'd welcome either.
Would you have the bibliographic references of the papers you've found?
My take on this is that it is of academic interest for those who listen to low-rez formats only. For those who fancy hi-rez these findings may have some value but then, without electronic x-overs and having the supersonic stuff in a separate channel you probably are listening to IM distortion. So are audio press and manufacturers selling misinformation, once again ? What about tests with western music and western listeners?
I would think the same (your point 1). Except that vinyl is not really (as) low-rez (as PCM). Doesn't it rather depend on contingent capabilities and properties of the vinyl rig what it can reproduce out of the analog source?
TIA for the Kaoru & Shogu paper. It could offer more fodder for thought on the significance of IM distortion in this respect (your point 2).
These won't qualify as adequate referencing but here we go:
- I stumbled upon a German tonmeister thesis work that I didn't read yet (bet your German is faster - ) but which seems relevant at least in parts. I got interested b/c of the two supervisors (Maillard and Meyer) who are pretty much the creme de la creme of the current-generation tonmeisters around.
- Griesinger's "Perception of mid frequency and high frequency intermodulation distortion in loudspeakers, and its relationship to high-definition audio" is a PPt presentation (at AES 2003 meeting) and I'll email it to you. Found ref to it in a NHK Labs paper by Nishiguchi at
Similarly only skimmed through a few pages so far. He is high credibility in the audio world as I understand it.
- The Boyk piece (mosty spectral analyses) I mentioned, "There's Life Above 20 Kilohertz! A Survey of Musical Instrument Spectra to 102.4 KHz," is at http://www.cco.caltech.edu/~boyk/spectra/spectra.htm.
Oohashi et al. are mentioned rather in passing in the (too) short discussion on significance. I couldn't find anything else by him, so I just wonder.
- And an item from HiFi News with some more references at the end (though none of these are very recent
No idea who this author is.
- Stereophile had a big editorial and discussion on this stuff at the early days of SACD & DVD-A, in 2000 or 2001 I recall (try searching "Atkinson Boyk").
- Best of all, I just realized that we may be rehashing a old debate from these very pages: see http://www.audioasylum.com/audio/general/messages/129397.html...
So it seems it's by no means a new discussion, even if nothing has really been established in the meantime. Indeed, why hasn't anyone done further studies on this?
If you want to test for audibility of sound above 20kHz. then you use test signals ONLY above 20kHz.
At best the study might prove that human hearing does not have a "brick wall filter" at 20kHz. But we already knew this.
If someone here wants to use a study to prove super tweeters are good components (for people who visit here), then he should cite a study using supertweeters playing ordinary Western music, not some bizarre "music" that contains an unusual amount of high frequencies, and then using brain measurements as a proxy for audibility.
Richard BassNut Greene
Subjective Audiophile 2007
Don't you think, RBNG, that brain scans would be better indicators of whether a difference was detected or not than asking someone, which can be full of bias from that listener?
The brain has a huge number of inputs, many of which are not noticed because they are masked by stronger simultaneous inputs. I think brain scans would not correlate with subjective sound quality.
If a blind listener, or listening panel, prefers the stereo system sound quality when a supertweeter is connected, versus when it is not connected, then you have a useful experiment to support buying that super tweeter for that stereo system.
We don't need brain scans ... and specially selected Indonesian music ... and only listeners under 35 years-old (whose hearing was never tested). That's too far removed from our reality to be a meaningful test.
"If you want to test for audibility of sound above 20kHz. then you use test signals ONLY above 20kHz."
Do you? There are 2 issues here. The first is whether or not we can hear pure toned above 20 kHz. For that, you test with signals only above that frequency. The second is whether HF content above 20 kHz affects what we hear below 20 kHz. For that you test with broad band signals as they did in this test.
"At best the study might prove that human hearing does not have a "brick wall filter" at 20kHz. But we already knew this."
Do we know this? How many people do you see here and elsewhere saying that we don't hear anything above 20 kHz and suggesting that supertweeters should therefore be inaudible. Many people believe "can't hear anything above 20 kHz" equates to saying we have a brick wall filter there. There's quite a deal of variation in what audiophiles appear to believe about this point based on what gets said here and elsewhere.
"If someone here wants to use a study to prove super tweeters are good components (for people who visit here), then he should cite a study using supertweeters playing ordinary Western music, not some bizarre "music" that contains an unusual amount of high frequencies, and then using brain measurements as a proxy for audibility."
I think you're wrong. We use test signals for determining what people can hear in the accepted audible range rather than "ordinary Western music" or speech, so why not use signals other than those for tests like this. What you want for test material is sounds with a lot of HF content and you'd probably want to vary the content below 20 kHz in order to determine whether frequency content below there affected how we respond to frequency content above there. The frequency distribution of sounds in Western music and in other musics such as gamelan music varies and there's no reason why Western music should be a more suitable 'test signal' than gamelan music. Since the study was done by Asian researchers, it's probably not surprising that they chose non-Western music for their tests. Western researchers are quite free to duplicate the study using Western music and that would probably be a good idea.
As for using brain measurements as a proxy, this does produce 2 useful results. First it removes the subjective response issue that would be present if the researchers only asked the subjects questions about what they thought they heard. You criticised such approaches quite strongly in your original "Dumb Study" post when you said "People will choose A or B because they are trying to please the experiment leader -- rather than saying "I don't know", or "I can't hear a difference". True of high end audio too!". Second, it relies on an objective physical measurement that the subject can't control so it produces quantifiable data which is always a plus in any study.
It does leave us with a problem about whether or not we wish to say that the person "hears" the hf content above 20 kHz if they don't report hearing anything when presented only with sounds above that frequency.
What we are left to deal with, if the results are repeatable and accepted, are a number of questions like:
- does the fact that we respond to hf content extending above 20 kHz affect our perception of that portion of the sound below 20 kHz? About all we can do hear is question the subjects and see whether they have any sort of repeatable, consistent response to sounds with extended hf content that is not present when they are presented with sounds with the above 20 kHz content filtered out.
- do people prefer sounds with or without such extended hf content.
To answer both those questions we're going to be forced to rely on the subjects' subjective responses to questions about their experience, but at least there can be available the objective data of each individual's brain responses as well and it may prove possible to correlate some of the subjective responses to particular aspects of the brain response.
There are probably some very good grounds for criticising this study, but I don't think your points here are among them.
There are many complaints about FM radio sound quality.
But complaints about output above 15kHz. seem rare.
When listening to typical Western music on FM radio, it seems that the
tiny amount of sound energy above 15Hz., assuming there is any,
can be eliminated and few people, if any, complain about its absense.
It seems to me that the people in the test said they preferred the music with content above 20kHz. WITHOUT ever proving they could hear a difference the two samples. A blind preference selection without the needed audibility test, is a typical high-end audio "blind test" (but not really a test at all).
There is also a possible question about how the energy above 20kHz was stripped from the music, and whether doing so created an audible artifact.
"There are many complaints about FM radio sound quality.
But complaints about output above 15kHz. seem rare."
What's to complain about? There basically isn't any :-)
Sorry, but you started out with this throwaway line and then threw it away anyway with your next sentence, so I couldn't resist this response.
"When listening to typical Western music on FM radio, it seems that the
tiny amount of sound energy above 15Hz., assuming there is any,
can be eliminated and few people, if any, complain about its absense."
You're right, but is that because our expectations of the quality of sound from FM radio are less than our expectations for quality from other sources? I enjoy listening to FM and I think it does a good job of doing what it does, but I wouldn't compare its sound quality to CD or vinyl. I listen to the radio for different reasons to those that I listen to my audio system for, and I both expect different things and accept a lower sound quality. I suspect most people concerned with sound quality are similar, even those who want to extract the most from FM as a source and who buy expensive tuners and antennae in order to get the most from it. I don't think your point here proves anything that helps your argument.
"It seems to me that the people in the test said they preferred the music with content above 20kHz. WITHOUT ever proving they could hear a difference the two samples. ""A blind preference selection without the needed audibility test, is a typical high-end audio "blind test" (but not really a test at all)."
It seems to me, from reading the table of the subjective results, that overall they thought that there was a difference on 5 parameters but that there was also not a significant difference in their like/dislike responses. The comment about preferring the sound with HF content is made in the text of the report but it did not seem to me to be borne out by the table of subjective results. That aspect of the report disturbed me. I think the authors assumed that if the subjects thought that the sound with hf content was perceived as "better" on those 5 parameters for which there was a significant difference, then the subjects preferred the sound with hf content. I don't think there was a strong preference for it because, if there was, then there would have been a statistically significant difference on the like/dislike parameter and there was not.
"A blind preference selection without the needed audibility test, is a typical high-end audio "blind test" (but not really a test at all)."
For a start we have the brain scans to show that there was a different response to the sounds with and without the high frequency content. There is also a statistically significant difference in subjective reports on 5 parameters. Audibility tests rely on the subject stating whether or not they hear the sound and that is a subjective report also. Determination of whether or not a sound is audible is done by whether or not there is a statistically significant response from a group of subjects, just as the determination of whether or not there were differences in the sound was made here. I think this is a valid test, but just because the test design is valid does not mean the results are valid. I think replication would be very good, especially replication with a variety of different sound types, before we started hanging too much on the results.
I'd also say that a "blind preference test", to use your terms (and I don't think the parameters on which there were significant differences in perceptions of the sound were preferences—I think the one parameter that did relate to preference failed to show a significant difference in perception) is valid. If people display a consistent and statistically significant preference for sound with hf content over sound without it, or vice versa, in a blind test, then it's plausible to believe that there is some attribute to the preferred sound that is linked to the preference. Whether or not that attribute is audible in isolation, ie the hf content on its own, is irrelevant. People may not be able to hear the hf content on its own but it may modify the way in which sound below the hf frequencies are perceived.
"There is also a possible question about how the energy above 20kHz was stripped from the music, and whether doing so created an audible artifact."
Agreed—that is an issue to be concerned about.
There are issues with this study which deserve further attention, and I'd love to see it replicated a few times with a wider variety of subjects and sound types before I hung too much on the results, but it is a much better quality study than you seem to believe.
The many test biases to "encourage" high frequency audibility:
-All the subjects were under 35 years old.
-Many were women.
-The Indonesian music had an unusual amount of high frequencies.
-Brain measurements were considered a proxy for "hearing"
-No one had to prove he/she could really hear a difference before stating a preference for music with and without the ultra high frequencies. (Beyond that complaint, the subjective evaluation conclusion didn't seem to follow from the data -- you covered that well in your post)
I just don't see how this study could support a decision to buy a super tweeter.
I mentioned FM radio in my prior post because there are no frequencies over 15kHz. ... meaning that a small amount of sound energy is eliminated for broadcasta (this didn't read as clearly as it should have) ... yet that fact NEVER seems to be mentioned among the many complaints about FM radio sound quality (even complaints by golden ears who think every wire sounds different).
That's a serious question.
IF (note: big "if") you want to find out whether HF content can be perceived by listeners in some way (note: I did not say 'heard' for reasons I will state a little later), then I think as an initial test it is more than reasonable to start with subjects who are under 35 and largely female, and with music with a high HF content, because if you can't get a positive result under those circumstances you would seriously need to consider giving up on the hypothesis. If you get a positive result under those circumstances, then you could be encouraged to continue with a wider range of test subjects including strong representation from older people and males, and with music with less HF content to see just how perceivable the HF content is for a wider and more representative set of subjects and with a wider range of music if, of course, you were interested in the question from a musical enjoyment angle rather than a straight perception angle which I'm not certain the researchers are. I think it's actually worth while starting out by seeing whether people who are most sensitive to higher frequencies in sound show a response than by starting out with a sample of people who are less sensitive to higher frequencies or even with a group of people representative of the population as a whole. I'd rather establish that some "more sensitive" people show a response and then see just how representative their response is than come up with a negative result from a less sensitive group that leads to the question being dropped and ending up not knowing that some people do show responses.
Now, as to "hearing". Our various senses register a wide range of stimuli including, in total, a much wider band of the radiation spectrum than any single sense registers. We can feel low frequency sound if the volume is high enough, and we can feel subsonic frequencies that we can't 'hear'. Does that feeling aspect of low frequency perception affect our experience of the low frequency sound that we hear, and does it affect our enjoyment of that sound? Under some conditions, I think it does. I like the vibratory effect I've perceived with bowed double bass sounds at time, and it adds to my enjoyment of the auditory aspect of the sound. Testing has failed to show that people can hear pure tones much above 20 kHz, and that we lose the ability to hear the last 5 kHz or so of that with increasing age. Does that mean that frequencies above 20 kHz don't contribute to our perception of sound? Not necessarily. What if frequencies above that are perceived by one of our other senses, as subsonic frequencies can be perceived? If that is the case, does such perception modify our overall perception of sound containing HF content in its overtone structure? That's a hard set of questions to answer but they are worth asking.
What this study attempts to do is to ascertain whether there is a difference in perceptual response to sounds containing HF content to similar sounds with the HF content filtered out. The brain measures are a good, objective means of determining whether there is a difference in response and this study indicates that there were differences. That's a good starting point but there are a lot of things that need to be sorted out before one can know what that really means. Amongst those things that need to be sorted out are:
- whether this study can be replicated with different subjects from the same demographic (ie under 35, largely female) with similar test signals. The reason for that is simply that this is the starting point and if there was some problem with this study that threw up a false positive result, it would be nice to find that out early;
- next, if replication is possible, what happens when the demographics of the test group are changed to a more representative sample? How do age and gender affect the results?
- next, vary the test signals with different sounds and music to determine how the proportion and nature of the HF content affects the results.
- and somewhere in the process, start trying to find out just what mechanisms are involved in the perception of the HF content. Is it hearing or some other sense.
- finally, at least as far as science is probably concerned but hopefully a little earlier for those of us with more personal interests in this particular question, is resolving the issue of whether such perception—once we're genuinely satisfied that it is present—modifies our enjoyment of music and in which ways does it do so.
I reckon this is a damn good study for an initial foray into an extremely interesting area. It doesn't answer all of the questions, and it raises a few of its own, but at least it gives some good strong support to a push for further investigation into the area and I think that's a very good thing. I think it was worth doing the study and I think it was worth publishing the results as they are because that may well encourage others to look at the topic and that's essential to the process of finding out a bit more about it.
As for your final comments on FM quality and sound quality in general. I don't know whether the best sound quality I've heard is from CD or from vinyl because I've heard damn good sound from both and we know vinyl does contain higher frequencies than CDs do. I quite enjoy FM radio from time to time. I'm also 60 this year and, though I haven't had my hearing tested in years, I do know that I can't hear a 15 kHz test tone on my system at the levels at which I normally listen to music. I also know that I prefer the sound of CDs on my system (no vinyl source in my system) to the sound of FM radio. Is that because CDs contain higher frequency content than FM—content at frequencies that I know I can't "hear", because my CD player is of better quality than my FM receiver, for some other reason, or for a combination of 2 or more of those reasons? I don't know the answer to those questions and I'd venture to suggest that neither do you. At least this study is a first step to answering some of those questions and it's nice to see that step taken.
I'm not unaware of problems with this study but I'd rather appreciate it for what it does than rubbish it because I don't like its shortcomings and/or have severe doubts about what it sets out to prove. I think the researchers deserve a fair bit of credit for their work on this one.
I think the study provides evidence that there is no logical reason to limit frequency response of any recording medium to 20kHz., even if no one could hear above 20kHz.
Unlike bass frequencies, treble over 20kHz. is:
(1) Not felt by the body
(2) Absorbed faily well by air in the room, and
(3) Easily masked by much higher music energy levels at other frequencies.
As soon as I noticed that the study used unusual music that contained an unusual amount of high frequency energy, I suspected the study leaders had an agenda.
Richard BassNut Greene
Subjective Audiophile 2007
"They are if the study is to correlate with people visiting this website and the typical music they listen to"
To paraphrase Carly Simon, "You're so vain, you probably think this study is about you".
Why should this study correlate with people visiting here and the music they listen to? For a start, science is rarely concerned with audiophiles and music lovers. We benefit from quite a bit of scientific research but that's a happy accident. I doubt Oohashi and his fellow authors have even heard of us, and that doesn't detract from the study one bit.
I think that really drags this thread to an end.
Give you a break because the study results were used by someone here to support the purchase of supertweeters by someone else here!
I'd be happy to give you a break if it weren't for one thing. Your first post was entitled "Dumb Study" and simply rubbished and, in my view, misrepresented some aspects of the study. I said I didn't think the study was so bad but all you've done since then is to try to trash it even further. You didn't about the use others made of the study at any time. You simply trashed the study at every opportunity.
I'd be happy to give you the break you asked for, for the reasons you want it but regardless of what others have said in this thread, my debate with you has never been about whether or not the study could be used to support the purchase of supertweeters. It's been about the general worth of the study, as a scientific study, right from my first response to you.
"IF (note: big "if") you want to find out whether HF content can be perceived by listeners in some way (note: I did not say 'heard' for reasons I will state a little later), then I think as an initial test it is more than reasonable to start with subjects who are under 35 and largely female, and with music with a high HF content, because if you can't get a positive result under those circumstances you would seriously need to consider giving up on the hypothesis."
This is simply starting at one extreme of possibilities to give the "best case" set of conditions. I see nothing scientifically invalid at all with those and I am not sure why RBNG thinks these are biases.
It also seems to me that brain activity is a much better indicator of whether someone heard a difference than simply asking them, which can be full of bias. Involuntary responses are most likely more accurate indicators.
As to FM, I am not sure why RBNG decided to muddy the waters with this comment. Since no one has ever heard FM with content ABOVE 15Khz then who is to say whether people would like it better or worse than with that content? Most people prefer cd or lp to FM radio in a high res system, is this because of more HF content?? Who knows.
Thanks for bothering to sum it all up.
Oohashi & co. did another study in 2006 in which their findings seem to suggest that this HF content perception "occurs" not through the auditory ear mechanism (air pressure through ear canal etc.) but through some other mechanism, as you surmise yourself, too. When listening was done through high-isolation earphones, no such "hypersonic effect" was shown. When the whole body of the listener was exposed to the signal (through loudspeakers), the effect was there.
For some time already there have been speculations (and more serious studies perhaps too) regarding perception of sound through bone conduction.
No one can prove anything.........
Chasing after "proof" in audio is a complete fools errand, - but I know someone around here who speaks so authoritatively, - one would think that he thinks that he's the most humble man that's ever walked the earth.... Heaven forbid that he actually goes out and gets some experience. But deep runs the fear of the man who is exposed with his pants around his ankles...
Well, hearing ultrasonic frequencies is controversial, to say the least. It takes more than just the signal being present and people being able to detect a difference. What seems to happen is that when when the ultrasonic frequencies can intermodulate with other frequencies in the speaker driver and produce intermodulation distortion (difference tones) in the audible range.
-Audio Engineering Society Convention Paper 5401, presented at the 110th Convention, 2001 May 12-15, Amsterdam, The Netherlands:
"Detection threshold for tones above 22 kHz," by Ashihara Kaoru and Kiryu Shogo, National Institute of Advanced Industrial Science and Technology 1-1-4 Umezono Tsukuba, Ibaraki 305-8568, Japan.
It seems as if the Oohashi studies are ok in this respect since they have used bi-channel systems with filter set at 26 kHz (170 dB/octave). However, audio systems as they are used in reviews and homes don't have electronic x-overs and multiamping. So, once again, the results obtained in listening tests using such systems are questionable, at best.
I am not at all certain Oohashi was able to remove the IM distortion, although they did make an effort. They say they eliminated the IM between the high frequency content and the lower frequency contentn by using two drivers. But since the IM would be difference tones, the tones in the speaker used for the high frequency content could very well have produced IM distortion products in the audible range. Measurements could have shown whether there were or not. Kaoru and Shogu showed graphs showing very clearly whether the IM products were present or not.
"Opposition brings concord. Out of discord comes the fairest harmony."
------Heraclitus of Ephesis (fl. 504-500 BC), trans. Wheelwright.
My best guess is that the behavior of the frequencies you can hear is actually changed by the presence of other frequencies. This is demonstrable at the other end of the spectrum. While ultrasonic high frequencies cause people to report improved midrange and sometimes improved bass, properly integrated infrasonic low bass seems to cause many to report improved midgrange and treble. In that case, since small waves ride big ones, assuming there is low frequency information in the recording, the presence or absence of those extreme frequencies, whether audible or not, will change the behavior of the sounds you CAN hear. As to whether the tiny ultrasonic waves will modify the bigger waves they sit on in a relevant way, I can't be sure. Haven't played with supertweeters much. But if it works for (good) subwoofers, it's probably true at the other end as well.
I thought the cutting head for LPs was 65khz? So ultra sonics have been with us for sometime.I dont know if its the ultrasonics that make these hifrequincy transducers sound good but it could be that many tweeters are rolling off in the frequincy ranges that humans still hear and the extended trebile that ultrasonic designs brings to the table is just extending this range. I find super tweeters etc to add a wee bit more air but if they are running in frequincys that humans hear they add more trebile energy.
So, in other words, ruler flat response to the hearing limit (and beyond) is what makes it sound "better," or distinctly "airy," "delicate," "spacious," "refined"?
I know it's possible to sense subsonic frequencies (you can even use them for torture or for creating panic, uncontrollable anxiety, and the like), but I just don't know if the same is true about ultrasonic frequencies and if so, whehter the effect then would be the reverse (inducing happiness in the "hearer"...).
This post is made possible by the generous support of people like you and our sponsors: