That's a great story about the bass clarinet, there is only one problem with it.
When you record the clarinet you also record the 6 kHz tone.
The more you think about these things the crazier it becomes. Especially psychoacoustics are really weird.
I read that Mr.Bose (yeah that one) stated that you can't distinguish between a 7 kHz sine wave and a 7 kHz square wave, that sort of things make you think. It does matches my experience with cheap piezo tweeters, if you filter them high enough they sound remarkably decent. Crazier still, i was reading a Hobby-Hifi and it boldly stated that the distortion of a certain loudspeaker added a nice sparkling brightness to the music. Even on this forum it's openly discussed how for instance the new Alpair 10 has more highs and how most people love those extra highs because otherwise the speaker sounds dull.
Other example: Tannoy put a "supertweeter" on top of it's loudspeakers, why would you go to all this trouble building a coaxial speaker and then ruin it? It seems that for high frequencies (>14kHz ?) it doesn't matter much where you put the tweeter.
The human hearing is limited by personal preferences. I believe that true hifi is an acquired taste.
I have just found this topic, and read Kunchur’s work with great interest – and some difficulty. His papers are summarised by George Foster in an article originally published in HiFiCritic – the no-advertising magazine published by Martin Colloms and edited by Paul Messenger. Kunchur’s website includes a link to this article, which receives his approval as a reasonable summary.
Some comments:
In “Temporal resolution of hearing probed by bandwidth restriction” Kunchur says he found that he couldn’t reproduce or measure small time differences in order to measure perception because he could not find digital test equipment accurate enough. So he used an analogue signal generator and built a high end analogue design headphone amp which could. He makes the point that earlier findings in this area of research may in fact reflect the shortcomings of the test equipment used. This puts a very big question mark over all the earlier research.
His findings do explain a lot of experiences some of us have with high resolution playback, ribbons and supertweeters.
As Foster points out, it isn’t just the ears, it’s the brain that does the work when we listen and it is capable of things far beyond our present means of measurement. Brain scanners don’t measure brain activity accurately, they only show where it takes place and its approximate magnitude.
Godfrey has posted a link to a paper by Oohashi et al about high frequency sound
http://www.linearaudio.nl/Documents/high freq inpact on brain.pdf
One of Oohashi’s findings was that even though the subjects could not consciously distinguish between full range sound (ie with frequencies above 22k) and sound with a 22k cut off, they reported greater enjoyment “Subjects felt that FRS was softer, more reverberant, with a better balance of instruments, more comfortable to the ears, and richer in nuance” (They used specially recorded Gamelan music),
Using brain scanners of various kinds, he also found inexplicably that full range reproduction of music caused more brain activity than when the 22k cut-off was used and some of it was in the parts of the brain that process pleasurable experience.
His research is over a decade old and Neurosciences have made enormous advances in identifying the roles of various brain areas – mapping the brain. The activity he found has been also related to listening to music by Zatorre (McGill University) Menon (Stamford), Levitin (McGill) and others.
Further Japanese research into high res by Nishiguchi: “Perceptual discrimination of very high frequency components in wide frequency range musical sound” in 2009 is not available free yet but there’s an abstract on the AES website AES E-Library: Perceptual Discrimination between Musical Sounds with and without Very High Frequency Components
It was overall inconclusive: it found that using tones out of 36 people tested “”two subjects, who could not hear the pure tone above 22 kHz, perceived differences between audio signals with and without a higher frequency band above 22 kHz only for a longer stimulus with the highest level of very high frequency components. We have no hypothesis or scientific reason that can explain this finding.” (The longer stimulus was a piece of music)
When we listen to music with an extended frequency range something physiological happens in some brains which we have difficulty in testing but are able to observe. In the absence of any other explanation Kunchur’s seems a pretty good one to me., and I’ll bet that the members of forums like this contain a higher percentage of those whose brains do react to the stimuli.
Some comments:
In “Temporal resolution of hearing probed by bandwidth restriction” Kunchur says he found that he couldn’t reproduce or measure small time differences in order to measure perception because he could not find digital test equipment accurate enough. So he used an analogue signal generator and built a high end analogue design headphone amp which could. He makes the point that earlier findings in this area of research may in fact reflect the shortcomings of the test equipment used. This puts a very big question mark over all the earlier research.
His findings do explain a lot of experiences some of us have with high resolution playback, ribbons and supertweeters.
As Foster points out, it isn’t just the ears, it’s the brain that does the work when we listen and it is capable of things far beyond our present means of measurement. Brain scanners don’t measure brain activity accurately, they only show where it takes place and its approximate magnitude.
Godfrey has posted a link to a paper by Oohashi et al about high frequency sound
http://www.linearaudio.nl/Documents/high freq inpact on brain.pdf
One of Oohashi’s findings was that even though the subjects could not consciously distinguish between full range sound (ie with frequencies above 22k) and sound with a 22k cut off, they reported greater enjoyment “Subjects felt that FRS was softer, more reverberant, with a better balance of instruments, more comfortable to the ears, and richer in nuance” (They used specially recorded Gamelan music),
Using brain scanners of various kinds, he also found inexplicably that full range reproduction of music caused more brain activity than when the 22k cut-off was used and some of it was in the parts of the brain that process pleasurable experience.
His research is over a decade old and Neurosciences have made enormous advances in identifying the roles of various brain areas – mapping the brain. The activity he found has been also related to listening to music by Zatorre (McGill University) Menon (Stamford), Levitin (McGill) and others.
Further Japanese research into high res by Nishiguchi: “Perceptual discrimination of very high frequency components in wide frequency range musical sound” in 2009 is not available free yet but there’s an abstract on the AES website AES E-Library: Perceptual Discrimination between Musical Sounds with and without Very High Frequency Components
It was overall inconclusive: it found that using tones out of 36 people tested “”two subjects, who could not hear the pure tone above 22 kHz, perceived differences between audio signals with and without a higher frequency band above 22 kHz only for a longer stimulus with the highest level of very high frequency components. We have no hypothesis or scientific reason that can explain this finding.” (The longer stimulus was a piece of music)
When we listen to music with an extended frequency range something physiological happens in some brains which we have difficulty in testing but are able to observe. In the absence of any other explanation Kunchur’s seems a pretty good one to me., and I’ll bet that the members of forums like this contain a higher percentage of those whose brains do react to the stimuli.
Hi Nardis,
Interesting comments and thanks for going the great lengths to paraphrase your cited research. Sadly, the problems of verification and reliability remain. It's not possible to rely on this part of Kunchur's efforts. Its authoritatively flawed due to the inadequate number of control and test subjects. From what I've seen of the other related research you've quoted, they have inadequate numbers of control and test subjects, or their test environments weren't isolated.
The problems of research into the limits of human hearing should be solved. It is possible to come close to a definitive answer to the limits of human perception of high frequencies. It is not so much a technical problem, but one of scale and finance. The techniques of blind trails with a large number of control and test human subjects in controlled environments is well understood and would yield more reliable results. Such trials are repeatable and therefore verifiable.
The basis for reputable applied science is very clear:
For any research to be authoritative and deemed reliable, it must go through the process of authentication by other reputable institutions, the results of the original and repeated research then being published in a recognised journal..
Its sad to see an increasing number of research academics "by-passing" such processes by publishing their work in the wider world. Academic funding is increasingly being tied to publication. The pressure to publish and promote non-verified research is increasing as academic's compete for funding. Indeed, for many academic's to remain in their establishment positions, their publication rates and conference address work count towards their academic survival.
It's wise to pay particular attention the sample sizes used, methodologies and to processes of verification. I don't want to discourage you but your conclusions lack validity unless you can cite more authoritative verified research published in a recognised journal.
Since I'm busy extending the F range of drivers, I'd welcome the publication of authoritative research that clearly demonstrates a human ability to perceive audio signals above 20-kHz. Such outcomes may help me to better market my drivers.
Cheers
Mark.
Last edited:
What it does prove is that 6 out of 6 people in his study can hear that kind of difference. It means, authoratively that some people will benefit from extended high frequency extension.
So you cannot say that no-one will benefit.
Because of the small sample it cannot be extended to the whole population. but coupled with a growing body of other data it is at least indicative. We now know that some of the population can.
dave
Hi Dave,
You've surprised me with the rather "direct" last post. I'm not sure why you've decided to attack me for arguing the case for more reliable larger scale research. I find it offensive when someone tells me what "I cannot" say.
6 people in what amounts to an uncontrolled test won't be accepted as "authoritative" by any robust research body or reputable journal.
It's not good practice to rely on research that uses such small sample sizes. There's no way to properly account for the variables within such small sample groups. It's not wise to "automatically" accept research that's not been repeated and validated by other institutions.
I can't agree with the conclusions made in your last post without seeing the cited research in this thread independently repeated on a greater scale, validated and published.
I understand your personal desire for the notion of extended human hearing. I appreciate your work on Enabling. I've encouraged you to engage in independent evaluation under controlled conditions in order to better appreciate the changes being produced on the drivers. My worry for you and some others in the audio world is the potential to loose one's objectivity. I constantly check myself when designing and making drivers. My personal desire to extend full range driver performance has to be balanced against the needs of system builders. Not an easy task and explains the reason why I like being part of the DIY forum. Would I like to see the case for greater human hearing abilities proven? Yes absolutely, but not by the shortcuts of inadequate non-validated research.
Mark.
Last edited:
Not an attack at all... just pointing out that the results are real, but, as you say, certainly not extensible to the entire population because of the small sample size. Within its limitations in numbers it was well conducted, at least the many referees for the 3 accredited journals that published the 2 studies, felt it worthwhile. An independent researcher has, it is noted, produced similar results. I have not seen that study.
As well Nardis points to some other studies (i have seen 2), that while not supporting the same exact capabilities as Kunchur studied, support the premis of detection beyond what has normally been considered the limit.
In audio circles there are always arguments about what people can and can't hear, and this study clearly shows that some people have time resolution capabilities this fine.
dave
I may have been a bit short as i've been trying to get a hold of you all day about an urgent matter that needs your attention, even tried to phone but the number i could find is invalid.
As well Nardis points to some other studies (i have seen 2), that while not supporting the same exact capabilities as Kunchur studied, support the premis of detection beyond what has normally been considered the limit.
In audio circles there are always arguments about what people can and can't hear, and this study clearly shows that some people have time resolution capabilities this fine.
dave
I may have been a bit short as i've been trying to get a hold of you all day about an urgent matter that needs your attention, even tried to phone but the number i could find is invalid.
First I need to apologise to Dave, I got a bit tetchy when I know he means well.
I just feel frustrated because as much as anyone, I'd love to see more investment of time and resources into human hearing research and the associated acoustical properties. I believe there's knowledge treasures in the acoustic world yet to be uncovered.
Re covers. I need to do a post soonest possible on this one. Will be glad of your input. I'll try to run the post as a poll, not quite sure how to do it but will try. Norman has suggested doing a new thread in Full Range. I think this is a good idea.
Cheers
Mark.
I just feel frustrated because as much as anyone, I'd love to see more investment of time and resources into human hearing research and the associated acoustical properties. I believe there's knowledge treasures in the acoustic world yet to be uncovered.
Re covers. I need to do a post soonest possible on this one. Will be glad of your input. I'll try to run the post as a poll, not quite sure how to do it but will try. Norman has suggested doing a new thread in Full Range. I think this is a good idea.
Cheers
Mark.
Last edited:
Take a look at this new thread:
http://www.diyaudio.com/forums/full-range/162321-alpair-10-cover-not-cover-question.html
Cheers
Mark.
Kunchur has been debated on many forums last year.
My position is this: assuming that the experiments are valid and that people regurarly can distinguish between two sounds that mutually differ only in a 5 us feature, even then is K's conclusion, namely that CD has an insufficient sample rate for conveying this, *wrong*.
The reason is that thanks to the blurring effect of the mandatory anti-alias filter at recording the 5 us distinguishing feature is spread out and allows the listener to hear a difference between the two sounds: yes, two impulses spaced 5 us apart would be destroyed, but what comes out of the channel would be different, and sound different, from two impulses spaced 10 us, or two coincident impulses.
The temporal or spacial accuracy of a sampled system like CD is much better than the inter-sample period. This is a fact that is exploited in other systems ranging from cellphones to optical spacecraft attitide sensors.
My position is this: assuming that the experiments are valid and that people regurarly can distinguish between two sounds that mutually differ only in a 5 us feature, even then is K's conclusion, namely that CD has an insufficient sample rate for conveying this, *wrong*.
The reason is that thanks to the blurring effect of the mandatory anti-alias filter at recording the 5 us distinguishing feature is spread out and allows the listener to hear a difference between the two sounds: yes, two impulses spaced 5 us apart would be destroyed, but what comes out of the channel would be different, and sound different, from two impulses spaced 10 us, or two coincident impulses.
The temporal or spacial accuracy of a sampled system like CD is much better than the inter-sample period. This is a fact that is exploited in other systems ranging from cellphones to optical spacecraft attitide sensors.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.