My hearing goes to 16kHz on sine waves, still i can apreciate a tweeter that goes to 30kHz -3dB. I my measurements i found that "fast" tweeters perform better in the time domain. For example the "waterfall" looks cleaner even in band. Resonance higher up seem to "fold down" into the audible range. I also now about research done in Japan with older people that got exposed to high resolution audio with high sampling rate. When the response in the treble was truncated they heard a difference and the brain wave detector sampled a difference. One explanation is that higher frequency signals modulate down into the audible range, for example a 22kHz and 23kHz signal played simultanuisly over a nonlinear transduces like a loadspeaker makes an audible 1kHz intermodulation product 23kHz - 22KHz = 1kHz
Yes, but if you wanted to produce two impulses of a given amplitude 5ms apart, you could consider that to be equivalent to a section of a square wave tone of 100hz. So you'd need a transducer capable of 100hz to give even a crude reproduction of such a pair of impulses, and a transducer capable of several times that to reproduce it as an accurate square wave.
Now, assuming it's true that we can percieve impulses at a 5us resolution, you need a transducer capable of frequences >> 100 KHz to produce such impulses.
Hopefully my square wave analogy illustrates why you don't need to think of it as a tone.
OK, just trying to get my head around this.
Lets say you have a woofer that is reproducing a 30Hz tone. The tone ends, the signal goes to zero. Then 5 micro-seconds later a 40Hz tone is fed into the transducer. Is the theory that the woofer must be able to reproduce frequencies > 100kHz to be able to reproduce the transition from 30Hz to 40Hz correctly?
Can't say I get it. The signal going to zero means the transducer has nothing to do. So why does it have to be able to reproduce 100kHz to then handle the second (40Hz) signal? Maybe I'm missing something - happy for someone to steer me onto the right way of thinking about this.
Because it needs to be capable of physically moving a certain distance (corresponding to the amplitude) in a particular time, ie. the rise time of the signal. In short, it needs to be fast, and that in turn corresponds with the ability to reproduce high frequencies.
(this is probably why subwoofers with extended high frequency response are thought to sound "faster" - they're better able to reproduce signals which aren't particularly sine-wavey. In the extreme case, a square wave can be seen to be made up of a fundamental frequency and the summation of an infinite series of harmonics)
(this is probably why subwoofers with extended high frequency response are thought to sound "faster" - they're better able to reproduce signals which aren't particularly sine-wavey. In the extreme case, a square wave can be seen to be made up of a fundamental frequency and the summation of an infinite series of harmonics)
Ok, so in this example the transducer needs to stop moving when the 30Hz signal goes to zero. Then start moving again on receipt of the leading edge of the 40Hz tone. And do both within 5 microseconds.
So what you are saying is (I think) that the ability to stop and start within such a time entails that the transducer could reproduce a 100kHz tone, if given one. Therefore, conversely, that if it can't reproduce such a 100kHz tone then it wouldn't be able to correctly reproduce signals with small time gaps.
I suppose its a bit of a moot point anyway, given that the analogue signals fed to speakers are continuous in time. And of course woofers by design aren't great at reproducing 100kHz signals.
That's basically my argument, yeah. That said, I'm far from convinced that any such "time detection" mechanism:
a) Exists
b) Makes any difference to our perception of music
I do believe, as Mark does, that having tweeters extend into the 30khz range will, as a direct result of the improvements necessary to do so, result in greater fidelity in the audio range. This is for reasons similar to my "fast" subwoofer hypothesis above.
a) Exists
b) Makes any difference to our perception of music
I do believe, as Mark does, that having tweeters extend into the 30khz range will, as a direct result of the improvements necessary to do so, result in greater fidelity in the audio range. This is for reasons similar to my "fast" subwoofer hypothesis above.
Wince. Controvertial subject if ever there was one (shades of speaker / interconnect wires etc). Not being the brightest, my own observations have to be fairly clunky, & far from original, but FWIW:
1/ I'm in complete agreement with the idea that the work necessary to increase the HF bandwidth of a driver (higher quality materials, tighter manufacturing tolerences &c.) might result in improvements lower down. It may also push non-linearities up, out of the main audio-band. But
2/ This isn't a given; it depends on the quality of the engineering. Sadly, some companies do it purely in bout of 'numbers oneupmanship.' Mark doesn't, but the same doesn't necessarily hold true elsewhere.
3/ Not everyone actually benefits from massive HF extension; you might get an improvement in non-linearities lower down, but if it comes at the price of, say, more HF energy than you personally like / enjoy, then it's not really done you any favours.
4/ Massive HF extension has to be context dependent. If we're talking wide-band drivers here, a large one is going to be beaming something chronic in the extreme HF. You might get some of the potential benefits lower down, but assuming for the sake of argument (without in any way conceeding the point either way), there is some benefit purely to having, say, extension an octave above our notional HF limit, it's surely going to be extremely limited in that respect.
5/ Most of the dominant storage formats are BW limited to quite a low frequency, and if it's not on the recording, it's not going to be reproduced by the speaker. QED.
6/ Kunchur's papers make very interesting reading & it looks to me to be a field worth further exploration. I'll certainly be keeping an eye on future developments, but at present, they still appear to be largely theoretical initial studies with very small sample groups. Seem to indicate some interesting trends, but I'll probably reserve judgement until they've had chance to do a larger scale study.
YMMV of course.
1/ I'm in complete agreement with the idea that the work necessary to increase the HF bandwidth of a driver (higher quality materials, tighter manufacturing tolerences &c.) might result in improvements lower down. It may also push non-linearities up, out of the main audio-band. But
2/ This isn't a given; it depends on the quality of the engineering. Sadly, some companies do it purely in bout of 'numbers oneupmanship.' Mark doesn't, but the same doesn't necessarily hold true elsewhere.
3/ Not everyone actually benefits from massive HF extension; you might get an improvement in non-linearities lower down, but if it comes at the price of, say, more HF energy than you personally like / enjoy, then it's not really done you any favours.
4/ Massive HF extension has to be context dependent. If we're talking wide-band drivers here, a large one is going to be beaming something chronic in the extreme HF. You might get some of the potential benefits lower down, but assuming for the sake of argument (without in any way conceeding the point either way), there is some benefit purely to having, say, extension an octave above our notional HF limit, it's surely going to be extremely limited in that respect.
5/ Most of the dominant storage formats are BW limited to quite a low frequency, and if it's not on the recording, it's not going to be reproduced by the speaker. QED.
6/ Kunchur's papers make very interesting reading & it looks to me to be a field worth further exploration. I'll certainly be keeping an eye on future developments, but at present, they still appear to be largely theoretical initial studies with very small sample groups. Seem to indicate some interesting trends, but I'll probably reserve judgement until they've had chance to do a larger scale study.
YMMV of course.
Last edited:
Hi Guys,
Excellent debate going on. I'm up early ploughing through emails and just caught this contribution from Scott. I gather that the number of human test subjects was very small, mentioning a total number of 22 humans, spilt into 5 groups of varying size.
This is an unrepresentative trail size. The variability risk is very high and therefore specific outcomes made by Kunchur should be considered "Unreliable". Sorry to disappoint but Kunchur's papers aren't "authoritative". Considerable care should be taken when using his work to support the argument for extended human hearing beyond 20-kHz.
At least his papers are making a contributing to the general debate. I think this maybe Kunchur's intention although it would be nice to see him specifically state this.
Cheers
Mark.
Last edited:
Hi Guys,
I've attached a gif pic that maybe of interest. Brian from Diyhifisupply sent me this graph a few years back. Note the frequencies and harmonics for orchestral music instruments and vocals. None reach 20-kHz.
I don't have the background to how this data was measured or generated so should you know more about this work, please tell us more
I'm not advocating the argument that its not necessary to develop audio beyond 20-kHz. I'm posting this information to give folks an appreciation for some of the likely acoustic elements involved in generating music.
Cheers
Mark.
I've attached a gif pic that maybe of interest. Brian from Diyhifisupply sent me this graph a few years back. Note the frequencies and harmonics for orchestral music instruments and vocals. None reach 20-kHz.
I don't have the background to how this data was measured or generated so should you know more about this work, please tell us more
I'm not advocating the argument that its not necessary to develop audio beyond 20-kHz. I'm posting this information to give folks an appreciation for some of the likely acoustic elements involved in generating music.
Cheers
Mark.
Attachments
Last edited:
Hi Mark
I do agree with you. I am not advocating tweeters should not extend beyond 20kHz. Afterall, the power amplifiers that I design goes up to 200kHz as a minimum (small signal) for the simple reason that if I were to limit the bandwidth to 20kHz, they sound awful.
But to say that we hear beyond 20kHz is a different issue. It is my contention that we do not hear beyond 20kHz. Yet the extended bandwidth often sounds better. Clearly, there are other mechanism involved which we are still unaware of.
Regards
Mike
I do agree with you. I am not advocating tweeters should not extend beyond 20kHz. Afterall, the power amplifiers that I design goes up to 200kHz as a minimum (small signal) for the simple reason that if I were to limit the bandwidth to 20kHz, they sound awful.
But to say that we hear beyond 20kHz is a different issue. It is my contention that we do not hear beyond 20kHz. Yet the extended bandwidth often sounds better. Clearly, there are other mechanism involved which we are still unaware of.
Regards
Mike
Could this improvement in sound be down more to the design pushing the dome resonance higher up. I have consistantly found if you notch out the dome resonance on tweeters they sound better, even though it is inaudible, at least to me.
I have intentionally set it off to see if it produced audible tones lower in the tweeter and it can but only at fairly high volumes (on one sample tweeter, so no statistical power to this test at all!) (I used meaurement equipment to do this not my ears as the levels were high and I am unsure about the risk to hearing, probably didn't do the tweeter much good either).
Just a hypothisis no real data.
Regards,
Andrew
I have intentionally set it off to see if it produced audible tones lower in the tweeter and it can but only at fairly high volumes (on one sample tweeter, so no statistical power to this test at all!) (I used meaurement equipment to do this not my ears as the levels were high and I am unsure about the risk to hearing, probably didn't do the tweeter much good either).
Just a hypothisis no real data.
Regards,
Andrew
If you do a wide bandwidth CSD (waterfall) plot you will see that the tweeter resonance broadens in the time domain and after a wile apaers in the audible range if the breakup is lower then 30kHz. The tweeters i use have only a diameter of 19mm of the metal come so resonace is at over 30kHz and does not fold down into the audible range in the time domain. That tweeter does not sound metallic or spitty.
Hi Guys,
I received 4 emails during the past week asking me about my thinking on "can humans hear above 20-kHz". Ummmmmph............I'm open to the idea that humans may have extended hearing abilities provided more large scale research can demonstrate an outcome to this effect.
For Derek in the UK, I understand your argument that just because there's no current proof, it doesn't mean that humans don't have extended abilities. While I have some sympathy with this thinking, as an engineer I'm more in the camp that wants to see proper research done in order to better assess this issue. With enough scientific research effort, I believe the question of extended hearing could be better illuminated.
My concern with siting Kunchur's work in favour of the argument for extended hearing, is that this research isn't robust. As I've already stated, the limited number of human test subjects precludes it from being reliable. The theories and outcomes generated from this research may be useful in aiding further research; But only time will tell if Kunchur's work is enlarged, or is picked up by another institution.
Sorry if I offended anyone but hopefully this helps folks understand my current thinking.
Cheers
Mark.
I received 4 emails during the past week asking me about my thinking on "can humans hear above 20-kHz". Ummmmmph............I'm open to the idea that humans may have extended hearing abilities provided more large scale research can demonstrate an outcome to this effect.
For Derek in the UK, I understand your argument that just because there's no current proof, it doesn't mean that humans don't have extended abilities. While I have some sympathy with this thinking, as an engineer I'm more in the camp that wants to see proper research done in order to better assess this issue. With enough scientific research effort, I believe the question of extended hearing could be better illuminated.
My concern with siting Kunchur's work in favour of the argument for extended hearing, is that this research isn't robust. As I've already stated, the limited number of human test subjects precludes it from being reliable. The theories and outcomes generated from this research may be useful in aiding further research; But only time will tell if Kunchur's work is enlarged, or is picked up by another institution.
Sorry if I offended anyone but hopefully this helps folks understand my current thinking.
Cheers
Mark.
Last edited:
There are at least 2 studies that examined whether humans could detect extreme HF. I don't have the refs handy, but one was Japanese. They have been discussed somewhere in threads here.
dave
Quote from Allen Wright's "The SuperCables CookBook" (recommended Vacuum State - High End Hifi Equipment)
dave
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.