I guess my 57 year old girlfriend hearing 20 kHz puts her in the one in a million category (she certainly is to me) then.
We are within months of the same age, the last time we tested she can hear 20kHz at a lower SPL level than I hear 10 kHz.
Her mother needs hearing aids to hear normal conversation.
I wish I could still hear past 16 kHz, but wishing or amplification won't make it happen.
Life is not fair, people's hearing varies considerably, and your mileage may vary
.Art
I think it's a matter of audible change vs. intermittent annoyance with music.
A quick "A/B" test easily demonstrates the audible change, but it takes more time (at least the first time) to become more conscious of the narrow/higher Q annoyance.
Ironically, once you've learned the lower/broader Q deviation - you dismiss it (or basically do an extra bit of eq.). Same thing that happens with most reflections.
On the narrow/higher Q though, once you've identified it - it becomes increasingly intolerable. (..think of a low-level "clicking" noise occurring somewhere near you - takes a bit to figure out that it occurs, takes a bit longer to try and locate it, and after a bit you can't take it anymore.)
I wish I could still hear past 16 kHz, but wishing or amplification won't make it happen.
Life is not fair, people's hearing varies considerably, and your mileage may vary
Art
While many have hearing damage or loss as they age, I'd also suspect that a good number of those that are reported as having hearing loss (at only much higher freq.s) due to age are instead subconsciously filtering it out - particularly men.
Most of the stuff "up there" is just noise, annoying noise. I'd count yourself blessed (or perhaps more "advanced").
That's completely natural, and also useful. Your ears are different, and measure differently.
People get this "hearing" thing all mixed up. Let's use amplifier terminology here. We say that the "bandwidth" of an amplifiers is from -3 dB low to -3 dB high. But then we claim that our hearing bandwidth is what ever we can detect with little to no reference to levels. There is not one person that I have ever met, and I have seen a lot of audiograms, who has a bandwidth, by the amplifier definition, to 17 kHz. Everyone's hearing sensitivity drops above 8-10 kHz. Some faster than others but everyone's does. Simply detecting a signal is simply no quantification of anything. It could be sub-harmonics, it could still be 20 dB down and it is "detected". But that is not the point. The point is that once it is more than 10 dB down it no longer enters into the "perception" picture because it is simply masked by everything below that frequency.
This is why I completely discount any and all claims that above 10 kHz as perceptually important. In only a rare individual could this be detected and in almost every case real music would not be detectable. Its importance numbers simply fall to nothing. This is precisely why Perceptual coders just don't do anything above 10 kHz. Because in real testing they found that it made no difference.
Asymmetric hearing loss is not natural and it can be an indication of other more insidious problems. No natural phenomena will be asymmetric.
This is why I completely discount any and all claims that above 10 kHz as perceptually important. In only a rare individual could this be detected and in almost every case real music would not be detectable. Its importance numbers simply fall to nothing. This is precisely why Perceptual coders just don't do anything above 10 kHz. Because in real testing they found that it made no difference.
Asymmetric hearing loss is not natural and it can be an indication of other more insidious problems. No natural phenomena will be asymmetric.
Earl,
I would assume that a normal hearing curve follows the Flecther-Munson type curvature? But does this curve mean that we can't hear those high frequencies or that it is just our sensitivity that varies with frequency? I'm not sure it is only the lower harmonics that we hear, but that we may still detect the higher frequencies and get limited information that we use to identify certain sounds. I am thinking of the sheen of a small triangle or some of the splash from a cymbal that lets us tell one from another?
I would assume that a normal hearing curve follows the Flecther-Munson type curvature? But does this curve mean that we can't hear those high frequencies or that it is just our sensitivity that varies with frequency? I'm not sure it is only the lower harmonics that we hear, but that we may still detect the higher frequencies and get limited information that we use to identify certain sounds. I am thinking of the sheen of a small triangle or some of the splash from a cymbal that lets us tell one from another?
I had a resonance at around 20KHz+. Everything sounded fine until I listened to Mr. Domingo sing, and it was only in a few notes range where there was a certain sonic signature heard. After working on a solution to solve the breakup which speeded the decay rate, the issue was significantly improved, but not perfect. After amplifier circuit modification and relayout, the symptom was further reduced. Normally I could not hear above 16KHz when played alone on the same setup. EQ will help because you put in less energy at that frequency, but then you start to effect the transient onset. Works if you accept this kind of compromise.
Since a new amplifier with increased decoupling capability is in the works, I think it would be interesting to see how things go.
I guess my 57 year old girlfriend hearing 20 kHz puts her in the one in a million category (she certainly is to me) then.
We are within months of the same age, the last time we tested she can hear 20kHz at a lower SPL level than I hear 10 kHz.
Her mother needs hearing aids to hear normal conversation.
I wish I could still hear past 16 kHz, but wishing or amplification won't make it happen.
Life is not fair, people's hearing varies considerably, and your mileage may vary
Art
Earl deals with the statistics, we look for extraordinary talent.
Robh3606,
Thanks for your response. I would still like to hear what Earl says about the testing protocol since his wife is a licensed audiologist. I could ask my mother who is a nurse and gives hearing tests but I know that she wouldn't know the technical aspects of the actual test equipment or whether there is an eq curve embedded in the equipment. The Fletcher Munson curve changes with spl so that makes me wonder how the two relate to our hearing response.
Thanks for your response. I would still like to hear what Earl says about the testing protocol since his wife is a licensed audiologist. I could ask my mother who is a nurse and gives hearing tests but I know that she wouldn't know the technical aspects of the actual test equipment or whether there is an eq curve embedded in the equipment. The Fletcher Munson curve changes with spl so that makes me wonder how the two relate to our hearing response.
Hello Kindhornman
I am no expert but I would think there would be differences in the sense that Fletcher Mulson is a group of equal loundness curves at different SPL's. The audiogram just seems to give a range for "normal hearing" from -10 to 20db with no bias as far as frequency is concerned within that db band. I am curious as well to what Earl can add to the discussion.
Rob
I am no expert but I would think there would be differences in the sense that Fletcher Mulson is a group of equal loundness curves at different SPL's. The audiogram just seems to give a range for "normal hearing" from -10 to 20db with no bias as far as frequency is concerned within that db band. I am curious as well to what Earl can add to the discussion.
Rob
Attachments
Last edited:
I think you may need to be more specific. I'm a bit skeptical of anyone identifying a minor problem above 15k with any normal music signal, sure, but I also know if I boost everything from 15k up 10dB in a reasonably flat speaker I will think it sounds HORRIBLE. Should I be verifying whether this is my imagination? That's a test I can actually do.
"normal" hearing is defined relative to the zero dB Fletcher-Munson levels. So -10 dB HL (hearing Loss) means your threshold is 10 dB above the 0 dB F-M curve.
I am not saying that everything above 10 kHz is inaudible, although that is certainly true for me and far more of us than they are willing to admit. But in the big picture if you dropped out everything above 10 kHz and people didn't know that, virtually no one would complain or even take note. It is just not that important. There is very little to nothing left above 10 kHz in nature due to air absorption. Nothing from an orchestra reaches past the middle seats above 10 kHz. It's all absorbed. So why would our hearing ever have developed a sensitivity to frequencies in nature that do not exist? Only electronics can create these kinds of frequencies and present them for listening. And only marketing cares. It's all hype.
In tests at Bell labs many years ago people preferred a cut in response above 8 kHz. May have cut out some distortion, I don't know, but that's why telephones don't go any higher. You can completely remove above 8 kHz from speech with no measureable effect on intelligibility.
There are no standardized tests of hearing above 8 kHz to even make objective quantifications. That's why my wife says that you really can't even talk about it because you can't quantify it from a hearing perspective.
I am not saying that everything above 10 kHz is inaudible, although that is certainly true for me and far more of us than they are willing to admit. But in the big picture if you dropped out everything above 10 kHz and people didn't know that, virtually no one would complain or even take note. It is just not that important. There is very little to nothing left above 10 kHz in nature due to air absorption. Nothing from an orchestra reaches past the middle seats above 10 kHz. It's all absorbed. So why would our hearing ever have developed a sensitivity to frequencies in nature that do not exist? Only electronics can create these kinds of frequencies and present them for listening. And only marketing cares. It's all hype.
In tests at Bell labs many years ago people preferred a cut in response above 8 kHz. May have cut out some distortion, I don't know, but that's why telephones don't go any higher. You can completely remove above 8 kHz from speech with no measureable effect on intelligibility.
There are no standardized tests of hearing above 8 kHz to even make objective quantifications. That's why my wife says that you really can't even talk about it because you can't quantify it from a hearing perspective.
Asymmetric hearing loss is a bit different than asymmetric hearing.
fas42 didn't say it was the result of hearing loss.
Asymmetric hearing is completely normal. Not only is the ear shape different with measurably different results (often at very high freq.s), but perceptual processing is different as well. Ex. Hearing music is dominate in the left ear for most individuals, and speech from the right.
The strange thing is that it's the right ear that's more sensitive, and that's the one that complains first when I subject it to too loud a sound, for too long. And the latter is a behaviour that started when I was a teenager, when an idiot fellow pupil fired a starting pistol right next to that ear - so it should be the one to be showing its age ...
Interesting philosophy that above 10kHz is unimportant - I just fairly recently tried as an exercise, killing everything above certain frequencies, brickwalling music tracks at 8, 10, 12kHz, and playing back on cheap PC speakers: 8k is disasterous, 10 leaves the sound dead, flat, tedious to listen to, only at 12 was the sound essentially as original in 'spirit' - I think I would use 14k as a safety margin, personally ...
Interesting philosophy that above 10kHz is unimportant - I just fairly recently tried as an exercise, killing everything above certain frequencies, brickwalling music tracks at 8, 10, 12kHz, and playing back on cheap PC speakers: 8k is disasterous, 10 leaves the sound dead, flat, tedious to listen to, only at 12 was the sound essentially as original in 'spirit' - I think I would use 14k as a safety margin, personally ...
When you double the Q of a resonance the amplitude has to be increased by 3dB to be "equally audible" according to the research you site. Well guess what, doubling the Q of a mechanical resonance increases the amplitude by 6dB.
The higher Q resonance is less audible only if the amplitude remained the same such as is possible when electronically generating it with a PEQ, but increasing the Q of a mechanical resonance will always make the amplitude increase at a greater rate than that required to maintain equal audibility, so it becomes more audible.
This is why the often held belief that damping a mechanical resonance is a bad idea because the lower Q version of the resonance is less audible is false.
Also there is a difference between the nature of a low Q and a high Q resonance - a low Q resonance is perceived as an error in tonal balance, while a high Q resonance (at least at high frequencies) becomes a source of "harshness" which doesn't necessarily cause an apparent shift in overall tonal balance.
To me eliminating that harshness is more important than perfect tonal balance, thus high Q high frequency resonances must go...
Last edited:
It seems to me Bell Labs also found that for reproduction to have a fundamentally "normal" tonality, the product of the -3 dB points at the upper and lower limits of the bandpass should equal 400,000.
Thus a 20k upper limit is valid only if the lower limit is 20 Hz. If low end response is 3 dB down at 40 Hz, then the response should be 3 dB down by 10 kHz as well to sound "normal".
This also is why telephones used to sound pretty good with a working bandwidth of 200 - 2,000 Hz; the product is 400,000. I say used to, because I am not sure this same principle is still universally applied, judging by some of the phones I've heard recently.
I've heard of British audiophiles in the 70s employing bandwidth control boxes with a single knob smoothly opening up the window by simultaneously extending the upper and lower -3 points while keeping the aforementioned product at 400,000.
Being a British audiophile who started in the early '80s, I've never heard of that and I'm puzzled why that would produce a 'normal' tone. It would be interesting to see more on this. From my research on this before, what the Bell engineers in the 50s I think showed was that a telephone needed to pass c.200-3400 Hz for the purposes of speech intelligibility and without the bandwidth requirements becoming too expensive. Fricatives and labial sounds, however, go up to c.8kHz, but they can be cut off without perhaps too much being lost. I wouldn't draw too many conclusions as to the needed bandwidth for a speaker from any of this though. Dr Geddes - any ideas for where I could find the 'people preferred a cut at 8kHz' paper/work? I've looked and found nothing.
I often wonder about those so called tests where people preferred the high end cuts that are talked about. Perhaps it had to do with distortion from the high end component? Perhaps if you were listening to a full range speaker and it had a whizzer cone you would want to cut out that type of resonant high end? So many of the high end dome tweeters have terrible resonance about 15Khz, some even lower that is shown as just ragged response while what it really is is just resonances in the domes due to breakup modes. Titanium is notorious for this with high Q resonance. I once had a blind man walk into the room while using a high end inverted dome titanium tweeter and he identified it as titanium unsighted! I think this is one reason that Be diaphragms are many times preferred over aluminum or Ti as the first top end breakup mode is shifted higher.
You nailed it Scott. I couldn't agree more.
Broad low Q resonances are perceived as a tonal imbalance - something that we adapt to quite readily.A high Q high frequency resonance is not immediately obvious depending on the musical content - the narrower the resonance the less statistically likely it is to be excited by a sparse musical spectrum, especially at the top end. (Not all music has a sparse spectrum though!)
However when it is excited it can be highly irritating, and once heard and identified, cannot be unheard.
I'm not sure if much research has been done into WHY a high Q resonance can sound irritating. (adding "harshness" to the sound)
My own theory from empirical observation is that a lot of it is due to the steep amplitude slopes of a resonance causing severe amplitude modulation of any frequency modulated (eg vibrato) sound that happens to sweep across it.
If this is correct an implication of this is that not only would mechanical resonances add harshness at high frequencies, so too would anything that causes steep narrowband shifts in the amplitude response - including diffraction with a sufficiently short time delay to cause narrow band interference...
Put simply, response smoothness is more important than absolute flatness.
Last edited: