Upper Limits of Human Hearing

[tresch]

I have a curiosity that I'd like to enter in on this discussion, because the whole "20khz vs 30khz" issue intrigues me.

The logical side of me tells me that making a tweeter go to 30khz is pointless, as we can't hear that high, and in general, harmonic distortion applied to frequencies at the high end of the spectrum shoes up as even higher frequency harmonics, which are outside of hearing. Thus, the ability to reproduce a complicated wave at 17khz is irrelevant as all we're going to hear is a sin wave anyway.

The reasonable side of my brain knows that audio is complicated and I certainly can't predict the whole situation. I imagine the biggest advantage would be that having a tweeter that extends to 30khz might make for much more accurate transients in the higher frequencies which could make for some very subtle, but audible differences in sound quality. Perhaps, also, the same characteristics that allow a tweeter to play flat to 30khz are also beneficial when it comes to preventing harmonic distortions and other negative effects.

So, what I want to know is this: Have you tested multiple tweeters that are as close to identical as possible while offering different high frequency extension? As in, do you have a 20khz tweeter and a 30khz tweeter and have listened to them both? This could be fairly irrelevent, as getting the tweeter to go to 30 might necessitate such a broad redesign that the two are no longer comperable, but I figure if there was a way to test, being a designer/builder, you'd be able to put together a closer comparison than anyone else

 

[markaudio]

So, what I want to know is this: Have you tested multiple tweeters that are as close to identical as possible while offering different high frequency extension? As in, do you have a 20khz tweeter and a 30khz tweeter and have listened to them both? This could be fairly irrelevent, as getting the tweeter to go to 30 might necessitate such a broad redesign that the two are no longer comperable, but I figure if there was a way to test, being a designer/builder, you'd be able to put together a closer comparison than anyone else

Hi Tresch,
Very interesting points. Extending frequency range beyond the upper limits of human hearing has been a "hot" topic over the years and will no doubt continue. Take a look at some of the ribbon tweets that claim +40-kHz outputs!

Does extended range make any difference? Well, from my work on designing and making cones, the answer is yes to point. Extending the range of my cones has improved their resonance properties. Check out the various comments on the Alpair 7 and Alpair 12 and you'll pick up on the outcome of my work in this area. On the question of Tweet comparison, I end up making anything up to 12 pre-production variants (more sometimes) and test them before committing to production. I do compare them to other brands but I hope you'll appreciate my stand regarding not commenting on the performance of particular competitor products. More important, I'm involving more forum members into the testing process. This thread is part of this process.

However, I believe there is a limit beyond which it's more about "commercial hype" rather than genuine beneficial performance gain. Frankly speaking, taking any Full Range or tweets much beyond 30-kHz (ish) isn't likely to be beneficial. There's only so much data that can be extracted from an emitting surface. As yet, I've seen no authoritative and reliable tests demonstrating that humans can perceive sound above 20-kHz. I've read various "mush" claiming that humans can detect +20-kHz, but its mere conjecture, nothing reliably demonstrated and proven to acceptable scientific standards. Therefore in my case, I would say that extending the overall range of my cones has improved their output in the "hearing" human range.

Commercially, arrrrh, a different matter. Similar to the "pixel" growth race in digital cameras, audio makers try to out-do each other. A major focus in recent years has been increased range. For many buyers, especially those who aren't so concerned with the technicalities, they feel more comfortable buying into the extra performance. It's similar to car buying. How many of us own a car that can easily exceed the highest speed limits (allot). I haven't owned a car here in Hong Kong for nearly 10 years because its a small place with excellent public transport. But there are more high powered BMW's and Mercs here per Km of Hong Kong roadway that anywhere else. These cars rarely change up beyond second gear such is the traffic congestion! Consumption at almost any price is ingrained in much of human nature, sadly at the cost of our children's future. Sorry to be sounding negative but its the main reason I'm a believer in Full Range; Because it's about doing more with less. Equally pressing is I've got to pay the rent and Full Range is a very small market. So I'm trying to make an economical woof and tweet with the knowledge I've gained from Full Range.

Hope this helps.

Mark.

 

[planet10]

Very interesting points. Extending frequency range beyond the upper limits of human hearing has been a "hot" topic over the years and will no doubt continue. Take a look at some of the ribbon tweets that claim +40-kHz outputs!

Does extended range make any difference?

Kunchur has shown that humans can detect timing differences of <5 uS, which if you flip it over to FR is >40 KHz.

The ear does have seperate mechanism for time detection vrs FR detection, so one may have FR limited to <15k yet able to hear timing anomolies equivalent to >40 KHz bandwidth.

dave

 

[markaudio]

Kunchur has shown that humans can detect timing differences of <5 uS, which if you flip it over to FR is >40 KHz. The ear does have seperate mechanism for time detection vrs FR detection, so one may have FR limited to <15k yet able to hear timing anomolies equivalent to >40 KHz bandwidth.

dave

Hi Dave,
Please quote Kunchur's trial publication data, trail methodology, sample size and a list of other institutions that have verified his work in a science publication. If he's got it done to the point of acceptability, it will be in a science journal. What will matter is if his work agrees and verifies your "flip" statement.

Ear has a separate mechanism? Please quote the scientific sources that have trial outcomes that clearly demonstrate this mechanism allows humans to perceive frequencies to 40-kHz. Please include their published trials dates and the name of publication.

I could be convinced if there's a published large scale robust body of research from a recognised scientific institution that has:

1 - control and sample groups greater than 2000 subjects.
2 - verifiable test and data analysis methodologies.
3 - other institutions have repeated the research in order to authenticate the original.
4 - been published in recognised science journals.

I'd quite like to have my scepticism lifted. It would help to put me towards the front the extended frequency FR driver race.

I'm wondering if it may be useful to start a new thread on this issue? I expect its been done in other threads but it could be re-visited? I'm happy to do it my Markaudio section.

Mark

 

[gfiandy]

Hi, For anyone interested here;

Information for prospective students

http://www.physics.sc.edu/kunchur/papers/FAQs.pdf

Are his papers.

This site has a good discussion on some of the probelms with his work.

Interesting Papers re temporal resolution - Hydrogenaudio Forums

I am open minded but yet to be convinced.

Regards,
Andrew

 

[planet10]

Hi Dave,
Please quote Kunchur's trial publication data, trail methodology, sample size and a list of other institutions that have verified his work in a science publication. If he's got it done to the point of acceptability, it will be in a science journal. What will matter is if his work agrees and verifies your "flip" statement.

Andrew beat me to the link. 3 articles published in 3 different referreed (sp?) Journals, plus another researcher producing the same results. Pretty solid work to my mind.

The forum link was interesting. The same kind of threads you see on cables & tubes vrs transistors... strongly held beliefs being challenged can bring lots of discussion.

dave

 

[markaudio]

Hi Andrew,
By the way, many thanks for your contribution on baffle influence. Sorry I've not had time to investigate. I hope to in the anechoic chamber later next week.

Many thanks for posting the links to Kunchur. I'll take look over the next few days. Let's see if this research:

1 -has sufficient human control and test samples
2 -has an evidential relationship demonstrating human ability to perceive sounds > 20-kHz.
3- its trial work has been repeated and verified.

I think this is worth making a new thread and migrating these latest posts to it.

Thanks

Mark.

 

[markaudio]

upper limits of Human Hearing

Hi Guys,
I've made this new thread since this is a subject of sufficient merit. There's been other threads on this subject in past but it may be useful to continue the discussion here in the light of my work in extending the HF performance of full range drivers.

I've moved recent posts on this subject from:
http://www.diyaudio.com/forums/markaudio/154837-marks-proto-6-5-woofer-1-silk-tweet-combo-7.html

Thanks

Mark.

 

[c2cthomas]

Hi Andrew,
I hope to in the anechoic chamber later next week.

Mark.

Walking around in those things listening to my heart beat in my ears always made me feel a little queasy

 

[markaudio]

Agreed C2. It took me some time to become familiar and comfortable. It was very spooky the first time I walked into an anechoic chamber
Cheers
Mark.

 

[tresch]

Okay, so, I'm reading through this article:

http://www.physics.sc.edu/kunchur/p...isalignment-of-acoustic-signals---Kunchur.pdf

and I'm already having doubts. Granted, I haven't read the entire thing, and I will, because it's quite interesting regardless of how it applies to the topic at hand, but there seems to be a very large difference between what we are discussing and what he's trying to prove, than what he is actually testing.

What we are discussing is whether there are positive benefits to having a loudspeaker that has a larger bandwidth in the frequency spectrum than the limits of human hearing (specifically on the high frequency end of the spectrum)

Kunchur is attempting to prove that having a larger bandwidth IS helpful. His process is to test a human's ability to detect PHASE distortion by adjusting the relative distances between TWO loudspeakers playing the same signal.

This seems like judging apples by observing oranges, to me. If one is determining the audible affect of a SINGLE DRIVER's bandwidth in relation to human interpretation, one must do all testing using only a SINGLE DRIVER. The physics of a single transducer following a waveform are very different from the physics of two separate waveforms from separate sources combining at different times and producing a distorted signal due to the additive/subtractive nature of phase.

i.e. If you adjust the position of one speaker, the human detects an audible change not due to "time smearing," that occurs in a system above the audible frequency range, but because the audible frequencies are severely distorted through addition/subtraction in a way that has no comparison to the affects of a single driver system following a single waveform, at least as it applies to the ability of the human ear to detect it.

It is an interesting test that says a lot about driver placement and the like, but I fail to see how such a test would have any bearing on the discussion at hand.

Someone please chime in and let me know if I'm crazy or not.

 

[tresch]

To paraphrase myself, for potential clarity:

This test determines nothing about the applicability of a driver that performs above the human hearing range, it only determines the extent of the human's ability to detect phase addition/cancellation within the audible frequency range.

 

[planet10]

What his papers do, using 2 different techniques, is that the human ear + brain can detect time differences of less than 5 uS.

How you interpret how that affects hifi reproduction is still open to interpretation & further study.

It does give solid experimental evidence that suggests that limits set based on the ear + brains FR detection capability need re-examing and gives credence to people reporting that, say, adding a supertweeter, can be detected/heard despite reproducing above the persons HFreq limit.

It does give solid experimental evidence that suggests that not only FR but time response is important to consider.

dave

 

[markaudio]

Hi Guys
Alas, I'm heavily committed for the next week or so but it's nice to have a quick read of the new posts.

I well remember having a conversation some years back with Susan Parker. She's an expert amp maker based in the UK. Susan suggested that the issue may be more to do with audio component's ability to deliver audio data. i.e, the greater the F range of the equipment, the better the quality of music reproduction within the human hearing range.

Sadly I've go to dash but just thought it may be interesting to ask if anyone has references to any research done with this aspect in mind.

Cheers
Mark.

 

[boconnor]

It does give solid experimental evidence that suggests that limits set based on the ear + brains FR detection capability need re-examing and gives credence to people reporting that, say, adding a supertweeter, can be detected/heard despite reproducing above the persons HFreq limit.

I really don't understand your statement. There have been hundreds of thousands (and more?) of audiometry tests showing that humans can't hear above 20kHz.

You can easily test this: go to any competent audiologist and ask them to play tones (sine, square, triangle or whatever) through their calibrated headphones, starting at say 21kHz. There is absolutely no doubt - you will not be able to hear them.

How can adding a transducer that reproduces tones above 20kHz be of any audible benefit to a loudspeaker system?

 

[tresch]

Susan suggested that the issue may be more to do with audio component's ability to deliver audio data. i.e, the greater the F range of the equipment, the better the quality of music reproduction within the human hearing range.

It makes sense to me that this is where the biggest changes would come into play. Something along the lines of "the characteristics that help a tweeter play to 30khz are also helpful in other parts of audio reproduction"

But then, I dunno. Does making a woofer play to 5,000 help it more accurately produce a 1,000hz tone? In that case that seems like an obvious fallacy, or at least a huge oversimplification.

 

[TheSeekerr]

@boconnor : Because, according to that research, it's concievable that we can percieve two impulses which occur less than 5 microseconds apart, even if we can't actually "hear" them as a tone. To reproduce such impulses would require a transducer capable of 40khz+ in frequency terms, but I don't think anyone is suggesting that we'd actually be percieving melodic or harmonic tones, just some sort of additional timing data.

 

[planet10]

From Kunchurs paper there are 2 different mechanisms for detecting signal (he describes the nature of each in the paper). The 1st (and familiar) is what we measure as you described... FR. I'm good to 12-14k max. The 2nd measures time differences (and ages more slowly than the FR mechanism), and is the one that allows us to discriminate <5uS.

dave

 

[jcx]

What his papers do, using 2 different techniques, is that the human ear + brain can detect time differences of less than 5 uS.

How you interpret how that affects hifi reproduction is still open to interpretation & further study.

It does give solid experimental evidence that suggests that limits set based on the ear + brains FR detection capability need re-examing and gives credence to people reporting that, say, adding a supertweeter, can be detected/heard despite reproducing above the persons HFreq limit.

It does give solid experimental evidence that suggests that not only FR but time response is important to consider.

dave

people not understanding correlation keep confusing group delay resolution with bandwidth - they are not the same or even simply related


S/N dynamic range and observation/integration time are also required to estimate group delay resolution in addition to bandwidth - greater S/N and observation time can extend group delay resolution to far below 1/bandwidth


so the uS intra-aural delay results do not require any extended frequency response beyond the accepted 20-24 KHz limits to explain

 

[tresch]

@boconnor : Because, according to that research, it's concievable that we can percieve two impulses which occur less than 5 microseconds apart, even if we can't actually "hear" them as a tone. To reproduce such impulses would require a transducer capable of 40khz+ in frequency terms, but I don't think anyone is suggesting that we'd actually be percieving melodic or harmonic tones, just some sort of additional timing data.

Except when we're dealing with a single transducer, two impulses 5microsecds apart simply turn into a single wave of sound which contains sonic and supersonic information just like any other sound wave.

Given a single transducer, such timing issues seem irrelevant, because all sound produced by that transducer is relative to everything else the transducer is doing. A single transducer doesn't produce multiple waveforms that are offset at different timings, it just produces a single waveform which contains audio data, some of which may be above the range of human hearing.

The timing with which the driver creates the wave has nothing to do with it's ability to play high frequency waves. A driver is an analog device and will play an audio signal with whatever timing that it was fed to the driver, offset by a certain amount due to the driver's particular phase delays.

Another way to look at it is, say you have a subwoofer that only plays to 150hz, and you have it wired up through a digital sound processor. If you wanted to delay the output of the subwoofer by 5 microseconds in order to get it to match phase better with your woofers, would the subwoofer have to be capable of a 40khz signal in order to delay an extra 5ms? Absolutely not.

I should note I'm not trying to be inflammatory, I'm just trying to piece together the correlation between timing and frequency response in my head, and it's not happening, so I'm stating my thought process in hopes of either being clarified or to see if I'm thinking correctly.