If human's range of hearing is upto 20 khz...

Discussion in 'Computer Audio' started by RogueTadhg, May 23, 2017.

  1. RogueTadhg

    RogueTadhg Gawd

    Messages:
    997
    Joined:
    Dec 14, 2011
    A lot of the brands list their frequency responses way past 20 khz. Is this just marketing hype where all big numbers = better audio?

    I've looked online for human's hearing responses and we get to 20,000~ hz before it gets discomforting. So why list a number like 40 khz?
     
  2. EVIL-SCOTSMAN

    EVIL-SCOTSMAN 2[H]4U

    Messages:
    4,040
    Joined:
    Feb 28, 2006
    Tis a gimmick

    Most people cant even hear 20khz

    Fuk am at 14k
     
  3. Bandalo

    Bandalo [H]ardness Supreme

    Messages:
    4,246
    Joined:
    Dec 15, 2010
    Everyone's ears are different...the "normal" range for a person ends around 20kHz, but some people can hear higher, some can't.

    And when they list the response, they're indicating what their speaker/headphone can actually produce accurately. It'd be silly for them to DESIGN a speaker that's tuned for 40kHz, but nothing wrong with reporting what it actually does.

    Now if you're BUYING a speaker based on it's 40kHz performance, you're a dumbass.
     
  4. Astral Abyss

    Astral Abyss 2[H]4U

    Messages:
    2,615
    Joined:
    Jun 15, 2004
    My hearing cuts out after 14kHz. But honestly, there's not much in that range of sound anyway.
     
  5. GlacierNine

    GlacierNine Limp Gawd

    Messages:
    148
    Joined:
    Feb 9, 2017
    20KHz is "supposedly perfect" hearing, give or take a spoonful. That means no damage over time at all - essentially, it's the range that is audible to young children.

    For most people in their mid 20s or later, 14-17KHz is where their hearing tops out.



    The reason they quote more numbers is because...well, they can. Frequency responses are measured plus or minus 3 decibels. If your speaker is within 1db of reference at 20KHz, then chances are it'll be within 3db for quite a lot higher frequencies. Since it takes a few minutes more to test for that as well, they usually do, just so they can put it on the box.

    At the end of the day though, frequency response means essentially nothing in practical terms unless you're trying to match a subwoofer with some bookshelf speakers in a 2.1 system.
     
    Archaea likes this.
  6. B00nie

    B00nie [H]ardness Supreme

    Messages:
    6,204
    Joined:
    Nov 1, 2012
    They list the range past 20khz because distortion has 'modes' that extend an octave below and above the actual peak level. If your linear response goes high enough also these modes go beyond the hearing range.

    Also some people believe that somehow ultrasonic frequencies bring 'air' to the sound but it hasn't been scientifically explained yet.
     
  7. Pandur

    Pandur Limp Gawd

    Messages:
    239
    Joined:
    Apr 4, 2000
    Digital audio uses a sampling frequency of 44.1khz or higher because of the Nyquist theorem.

    The reason for this is basically that digital audio have a peculiar perk that causes alot of noise over half the sampling frequency. Ergo with a sampling frequency of say 22050hz (sticking with known computer values here) you would have alot of noise from 11025hz and up, and even older geezers can hear this. So it was pushed way higher to get the noise out of the audible range of any human.

    But beyond these ranges (44.1/48khz), we're into snake oil territory imho. I can accept 88.2/96khz but beyond that is starting to get really silly from a scientific point of view.
     
    SvenBent likes this.
  8. B00nie

    B00nie [H]ardness Supreme

    Messages:
    6,204
    Joined:
    Nov 1, 2012
    The Nyquist theorem is only valid if a perfect low pass filter is used. Unfortunately there is no such thing as a perfect filter available at the moment. So there is a technical point for using higher sampling frequencies.

    However considering that most adults can not hear frequencies above 15-16khz anyway, it's questionable if 44.1khz is even theoretically a problem. Certainly blind listening tests haven't been able to bring conclusive proof that it's a detectable problem.

    It all stems back to the same thing: The electronics is really a minor factor in the total chain in reality. The recording itself defines the major part of the quality, next comes the speakers. Sampling rates etc. play a very minute role.
     
  9. Pandur

    Pandur Limp Gawd

    Messages:
    239
    Joined:
    Apr 4, 2000
    It would be a really bad idea to make audio equipment that are producing horrible noise above the audible range of most >50 year old adults. You have to take the best case scenario as your baseline for this, and lucky for us the people in charge back when this was decided did just that. Which is one reason 44.1khz became the baseline. I'm sure the fact it's a doubling of the older standards of 22050hz, which in turn is the double of 11025hz also played it's fair part.

    If modern audio equipment had been designed by using the double of 15-16khz, we'd live in a world were most people under the age of 30 couldn't go near any consumer electronics that play audio. I'm sure that's somebodys fantasy... So let's at least give credit where credit is due.

    I linked to the Nyquist theorem to give the whole story. But to be exact; the sampling frequency is the double of your desired max audible frequency as a result of the Nyquist frequency part of the theorem. And I don't remember any requirement for a ideal low pass filter with regards to how this works. It's more of a byproduct of how digital audio sampling is done mathematically.

    Bottom line for me is I'd rather have a higher bit sampling rate (16 vs 24 bit) than sampling frequency, once said frequency is at least 44.1/48khz. And there is a sound mathematical reason why we have sampling frequencies in the 40k range instead of 20k. It's to save our young ones from headaches. :p
     
  10. B00nie

    B00nie [H]ardness Supreme

    Messages:
    6,204
    Joined:
    Nov 1, 2012
    You misunderstood me. The Nyquist theorem shows that a _perfectly_ lowpassed signal can be exactly reproduced using 44.1khz sampling. Due to the fact that there is no perfect lowpass filter that would cut the audio signal 100% sharply, artifacts result from the conversion of ultrasonic frequencies that exist at the normal hearing range. This is why higher sampling rates have been experimented with.

    Basically you either low pass the signal way below 20khz using existing crossover techniques or you get artifacts. Extremely steep crossovers bring their own problems with phase and delay.
     
  11. Tiberian

    Tiberian DILLIGAFuck

    Messages:
    4,739
    Joined:
    Feb 12, 2012
    Absalom likes this.
  12. GlacierNine

    GlacierNine Limp Gawd

    Messages:
    148
    Joined:
    Feb 9, 2017
    Last edited: May 26, 2017
    Absalom likes this.
  13. pippenainteasy

    pippenainteasy Limp Gawd

    Messages:
    257
    Joined:
    May 20, 2016
    Mostly just for advertising. Also without a +/- db range its not that useful of a performance metric. Technically your free earbuds go out to 40khz, its just output is probably down -50db.

    Its the same reason cheap boomboxes list their output in 1000s of watts. Technically true, if measuring instantaneous output at a specifically easy to amplify frquency (usually 1khz) at insane distortion levels. Not useful for comparing with any other products. It would be like if some cheap carmaker claimed their cars top speed was 200mph--tested while in an uncontrollable descent down Mount Everest. :-D
     
  14. Archaea

    Archaea [H]ardness Supreme

    Messages:
    6,909
    Joined:
    Oct 19, 2004
    Most people older than teenagers have lost hearing above say 16,000hz.

    There are free hearing test services online like http://www.audiocheck.net/ to establish your hearing if you have a good set of speakers or headphones.

    20000hz hearing capability is what we start with as a youngster. It wanes out as we age till you can't hear much above 8-12khz as a older person.

    40,000hz is pure marketing IMO.
    Even so...
    Every doubling of hz represents an octave. 40,000hz sounds super high and is generally above nearly anyone's human hearing, but it only represents one octave above 20,000 hz. An octave, in simplest layman terms is just eight major (white) notes on a piano. So it might not be as ridiculous as it sounds.

    I'm 37 and hear up to about 16,000hz and generally have above average hearing for my age according to a fairly recent audiologist visit.
     
    John721 likes this.
  15. B00nie

    B00nie [H]ardness Supreme

    Messages:
    6,204
    Joined:
    Nov 1, 2012
    The high frequency boom is purely about distortion products. The higher you can push the linear bandwith, the further the distortion products usually get from the listening field.

    A distortion product effects not only the exact frequency it inherently manifests itself but also one octave above and below it's natural range. With a 22khz distortion product, the 11khz reflection is already on most peoples hearing range.
     
    N4CR likes this.
  16. SvenBent

    SvenBent [H]ard|Gawd

    Messages:
    1,805
    Joined:
    Sep 13, 2008

    you have tho realise it take two samples to reconstruct a wave. So you need 44khz SAMPLES to reconstruct ~ 22Khz WAVES
    Don't mix samples and wave together.

    The 44.1KHz samplinh rate came from VHS as it was the leftover space on the tape.
    That standard followed over to CD
     
  17. Archaea

    Archaea [H]ardness Supreme

    Messages:
    6,909
    Joined:
    Oct 19, 2004
    I'm wasnt talking about samples. I spoke to frequencies in Hz.
     
    Last edited: May 27, 2017
  18. SvenBent

    SvenBent [H]ard|Gawd

    Messages:
    1,805
    Joined:
    Sep 13, 2008
    my apologies then.
    But those are both frequiences in hz though... but i assume you meant wave frequencies and not sample frequencies then.
     
  19. Sycraft

    Sycraft 2[H]4U

    Messages:
    4,002
    Joined:
    Nov 9, 2006
    As others have mentioned this is basically marketing BS. However there is one useful reason to care about this in some cases: Aliasing. If you take a sound that is higher frequency than a band-limited system can produce and feed it in to the system, you actually get audible artifacts called aliasing. You near about it the context of digital sound because it is something extremely important to deal with there but analogue systems can alias too. So how do you prevent the problem? Well you can just feed in a lower frequency signal, which is what we tend to do in the digital era. You can also build in filters to your gear to make sure it doesn't alias. However a third option is just make sure you have gear that can handle the higher frequency sounds. If they pass cleanly through then there's no aliasing, and no audible problems.

    Also some devices are simply going to have higher frequency response characteristics just because. Amps are that way. You could go through the trouble of designing an amp that cut off right at 20kHz but it would cost more, and sound worse. So you don't, you make an amp that'll sound good and it ends up having a bandwidth of a couple hundred kHz just because that's how the components work.
     
  20. N4CR

    N4CR 2[H]4U

    Messages:
    2,073
    Joined:
    Oct 17, 2011
    With speakers there is a reason for this.

    If you have the maximum reproduction frequency at 20KHz, it often is running in 'break up' or where the cone/transducer/dome/etc starts to distort as it approaches this zone.
    By moving this peak design frequency up the charts e.g. 40-60KHz, you have a cleaner high end reproduction and move the 'breakup zone' out of audible limits. You can see this with the B&W Diamond series of tweeters, it's measurable and quantifiable. The rigid diamond dome increases the 'break up' frequency of those tweeters.


    http://www.bowers-wilkins.com/Discover/Discover/Technologies/Diamond_Tweeters.html


    Edit: boonie explains this far more eloquently than I can and knows more about audio engineering than anyone on this board. This is coming from a guy who nearly started a subwoofer manufacturing company.​
     
    B00nie likes this.
  21. pippenainteasy

    pippenainteasy Limp Gawd

    Messages:
    257
    Joined:
    May 20, 2016
    Metal drivers aren't as well damped as soft domes. Soft domes, when they break up, do so locally, whereas metal drivers tend to break up across the entire driver, which is why you want the stiffest possible material when utilizing metal drivers. Soft domes perform well, but tend to have poor extension, generally rolling off from 17-18KHz on. This is even true for extremely high end soft domes like the Dynaudio Esotar2, which has an OEM price of somewhere around $1400/driver.

    As a result, metal ends up being a better bet if you are chasing extension. Beryllium is quite popular in the high end audio space, but it's quite a costly material, as vapor deposited beryllium costs by weight costs more than the spot price of gold or platinum.

    Some example diagrams I lifted from online of the advantages of a stiffer metallic driver, comparison being that of a 4" dome:

    [​IMG]

    [​IMG]
     
    Last edited: Jun 7, 2017
    N4CR likes this.
  22. westrock2000

    westrock2000 [H]ardForum Junkie

    Messages:
    9,607
    Joined:
    Jun 3, 2005
    A 4" dome at 20Khz though???

    Not exactly common place.
     
    N4CR likes this.
  23. B00nie

    B00nie [H]ardness Supreme

    Messages:
    6,204
    Joined:
    Nov 1, 2012
    For compression drivers it's normal.
     
  24. daglesj

    daglesj [H]ardness Supreme

    Messages:
    4,495
    Joined:
    May 7, 2005
  25. B00nie

    B00nie [H]ardness Supreme

    Messages:
    6,204
    Joined:
    Nov 1, 2012
    This is purely marketing for the dumb. You'd have to be completely oblivious of the basics of audio to be impressed with that ad.
     
  26. jlbenedict

    jlbenedict Gawd

    Messages:
    968
    Joined:
    May 22, 2005
    If human's can't hear about 14khz... is it "elitist" attitude then to think cd's/albums must be ripped in lossless format .. when 256k or 320k will do? Yeah.. we get it.. you have 100pb of storage.. but really though
     
  27. B00nie

    B00nie [H]ardness Supreme

    Messages:
    6,204
    Joined:
    Nov 1, 2012
    If you're talking about MP3 it doesn't limit the bandwith but it removes microinformation from the signal which humans are not supposed to hear. For example if you get a 1khz tone at -3db and there's a 800hz tone at -20db the mp3 filtering will simply cut it out since the ear will have a hard time hearing the lower signal under the loud one. Basically you get a whitewashed version of the original signal. To the ear the difference is very small but to the data processing it's like night and day.
     
  28. StoleMyOwnCar

    StoleMyOwnCar [H]ard|Gawd

    Messages:
    1,968
    Joined:
    Sep 30, 2013
    That Mosquito tone hurts my ears. It's not... "loud"... I guess ... but it's... it's just like it hurts in some other way. The modulating one is a bit more tolerable.

    Sigh, I've had tinnitus on and off for a very long time now, and sometimes I'd love to trade in these ears for ones that were less sensitive. @_@;

    I mean I've gone to a few rock concerts and still I can apparently hear that...
     
  29. Absalom

    Absalom Limp Gawd

    Messages:
    361
    Joined:
    Oct 3, 2007
    Using the audiocheck.net site as reference, I can't hear anything above 16kHz (at a reasonable volume that is). Kind of glad I can not. There's nothing pleasurable in that range anyway.

    I'm 42 years old, so I guess I've got some headroom before the inevitable hearing loss starts eating into frequency ranges I actually do care about.

    Also, the 320kbit mp3 is enough debate will go on forever. It's all context sensitive. There are some songs I can't tell the difference at all. And then there are others (usually metal or prog-rock) where I do notice a difference. Not everyone interprets aliasing artifacts the same. And some of it might be placebo effect.
     
    Last edited: Jun 19, 2017
  30. Khahhblaab

    Khahhblaab Limp Gawd

    Messages:
    225
    Joined:
    Apr 23, 2017
    One reason is that the extent of human hearing goes out to 20Khz so you have to design for that. Another reason is headroom. If you design circuitry to go to the average of 16khz, your frequency response will drop off wayyy before 16khz. So you wind up not getting a full 16khz. By designing circuitry to go out to 20khz, you guarantee( more or less) flat frequency response to 16khz. Another wierd reason is something that Harmon kardon discovered many years ago. There is some type of - not yet scientifically explained with out debate - interaction of frequencies at different levels. HK designed amplifiers that would go out to 200khz. Reviewers <most> have all stated how clean and clear their better amps are. HK amps uses these ultrasonic designs to be better at amplifing the human range. So designing for 20khz, may clean up sounds below 20khz. The circuitry that has to be able to 'do' 20khz, has an easier time doing 16khz. There may be other reasons but these are the ones that answer most concerns. 20khz betters response below 20khz and may provide a cleaner sound because if it.