[flargosa]

The Sennheiser HD800, Grado PS1000e, Beyerdynamic T1, and Hifiman HE560 all play up to 50 khz. Why design headphones to reproduce frequency ranges beyond 20khz if there is nothing there?
 

 

[Espresto]

Even though the fundamental notes may all fall under 5khz harmonics and overtones reach well above. This is especially apparent with cymbals and strings. If you cut out all the information above 10khz you'll notice an immediate difference in any recording of acoustic instruments. 
 
I do agree though that past 20-22khz it probably doesn't make a significant difference. I know loudspeakers reach far far higher though, so I may just lack experience.

 

[DaveK1977]

Also, if its range goes well outside of the limits of hearing, then that would suggest that it's not struggling to reproduce the frequencies that are at the very ends, in other words they're suggesting that its response curve isn't rapidly dropping at 18 kHz if the driver can actually reproduce 30 kHz.  A frequency graph would be more useful than just saying 5-30,000 Hz.  There's never any context of +/- x dB attached to those "specifications" either.  I would pay little attention to those unless it's quoting figures that are within the 20-20,000 range.  If its quoted range stops within those frequencies, then it will probably have a deficiency at that extreme that would be obvious to anybody.

 

[fauxfreshness]

  Also, if its range goes well outside of the limits of hearing, then that would suggest that it's not struggling to reproduce the frequencies that are at the very ends, in other words they're suggesting that its response curve isn't rapidly dropping at 18 kHz if the driver can actually reproduce 30 kHz.  A frequency graph would be more useful than just saying 5-30,000 Hz.  There's never any context of +/- x dB attached to those "specifications" either.  I would pay little attention to those unless it's quoting figures that are within the 20-20,000 range.  If its quoted range stops within those frequencies, then it will probably have a deficiency at that extreme that would be obvious to anybody.

 
It's much closer to this.  You'll find a lot of fundamentals are rather low in frequency, but the harmonics of those instruments and the extremely wide range of physical and virtual, electronic instruments (additive and subtractive synthesizers, for instance) can easily drive to either end.  I normally do a lot of "house keeping" along each instrument, and then each bus/group and tend to use a sharper roll-off (24 or 48 dB/octave) at the sub-bass end and a low-pass shelf at the high end to keep the effects (reverb, stereo imaging, and now binaural work) from being twisted up with a sharp cutoff right at 20 kHz and leave the resampling and dithering to mastering processes to get it from 24 bit/48 kHz to 16 bit/44.1 kHz.  When producing, even the deepest sampled instruments tend to be 24 bit/48 kHz, and the only instrument I've recorded directly is with the intent of distortion (electric guitar), so I monitor and calibrate beforehand anyway.  But hey, if you're recording it, throw whatever resource capacity you can at it as you never know what tech later on can expose details you were able to capture.
 
There are times where in the same house keeping process you might want to make sure the overall to and bottom ends don't get a chance to bleed over into the Nyquist cutoff, but I rarely have anything go beyond 22 kHz anyway.  A lot of times, depending on target format, the assumption is that the playback will likely mask the chop off at either end if using MP3 (depending on the CODEC and settings used) or some other lossy, possibly streaming format.  It's partly why I tend to research as much as possible before I bother getting "HD" versions of tracks.  And even then, I don't bother with anything over 24 bit/96 kHz until I can try DSD-targeted tracks (ones that were always intended to be mastered at high resolution).  Mostly because I just want to hear the difference for myself.  

 

[SP Wild]

Also, if its range goes well outside of the limits of hearing, then that would suggest that it's not struggling to reproduce the frequencies that are at the very ends, in other words they're suggesting that its response curve isn't rapidly dropping at 18 kHz if the driver can actually reproduce 30 kHz.  A frequency graph would be more useful than just saying 5-30,000 Hz.  There's never any context of +/- x dB attached to those "specifications" either.  I would pay little attention to those unless it's quoting figures that are within the 20-20,000 range.  If its quoted range stops within those frequencies, then it will probably have a deficiency at that extreme that would be obvious to anybody.

This.

I bought a pair of pro speakers with a horn tweeter. It was specced to 17khz within neutrality. That was all I need I believed as according to sine waves I can't even hear anything up there anyway.

The problem is it must be struggling to maintain 17khz to flat, what I hear in the treble is smeared, if not smooth...a horn trait is smoothness I believe.

Paper cone tweeters specced similarly sounded grained, and unresolved.

Dome and ribbon tweeters specced much higher have a better definition with clear sonic attack boundaries.

 

[conchchowder]

One word. Psychoacoustics.

 

[Mad Max]

I read some argument around here about "midband distortion" and how you want it up in the ultrasonics and not the audio band for best sound.
 
Then there's the matter of harmonics and all, but I doubt whatever's above 20k-22kHz matters.
 
It could just be that the headphones aren't necessarily designed to go up to 50kHz, but they just so happened to in the end.  I don't think it matters.
 
Edit: and Sony claims 100kHz for their SA5000, lol!

 

[ProtegeManiac]

  Also, if its range goes well outside of the limits of hearing, then that would suggest that it's not struggling to reproduce the frequencies that are at the very ends, in other words they're suggesting that its response curve isn't rapidly dropping at 18 kHz if the driver can actually reproduce 30 kHz.

 
This. Particularly with driver diaphragms that aren't dynamic types, the real reason why they sound a lot better up top isn't because they play well beyond the human hearing range, but that they are very smooth in the midrange and treble since it doesn't rely on excursion as much as dynamic drivers. The offset of course is that they have to rely more on surface area for bass, which is why you get more bass modules with wider surface areas for the Mini Maggies for example, or why some Martin Logans come with subwoofers mounted on the bottom.

Note though that this isn't always true whenever manufacturers quote very high frequency extension. Some may quote very high limits, but their dynamic drivers still aren't as smooth as planars. Some, like my ASG-1, can be rated beyond 20khz (25khz to be exact), but run a sine sweep through it and the treble rolls off at 12khz, and on normal listening levels, is barely audible until 16khz, and then disappears (even if I put it on full blast). By contrast the HD600 is rated up to 39khz, but on a sine sweep I can hear a little bit past 20khz (that or the generator screwed up where the shifts are) but zero at 25khz. Doesn't mean there's a lot of information past 16khz on what I'm actually listening to otherwise though.

 

[goldenSHK]

  Also, if its range goes well outside of the limits of hearing, then that would suggest that it's not struggling to reproduce the frequencies that are at the very ends, in other words they're suggesting that its response curve isn't rapidly dropping at 18 kHz if the driver can actually reproduce 30 kHz.  A frequency graph would be more useful than just saying 5-30,000 Hz.  There's never any context of +/- x dB attached to those "specifications" either.  I would pay little attention to those unless it's quoting figures that are within the 20-20,000 range.  If its quoted range stops within those frequencies, then it will probably have a deficiency at that extreme that would be obvious to anybody.

This is the right answer. Everyone already confirmed it, but I'll just +1 it again. 
 
I always notice that the bass is much smoother and truly hits way harder on headphones that have frequency response down to 5-6hz compared to 20hz which is standard on any headphone you find. The reason is not because you can hear or feel the other frequencies (in fact, most commercial DAC chips only convert 20-20 frequency response so unless you're using a high end $1,000+ DAC, you are only getting 20-20 response no matter what regardless if your headphone can produce more!!). The reason for such amazing and juicy bass is probably because most headphones that only go down to 20hz are dying to produce the ~20hz notes while the high end headphones that go down to 6hz for example can much more naturally and effortlessly attack those low tones. 
 
It should be noted that there are certain headphones that this principle doesn't really apply to much either such as the Alpha Dogs which has a frequency response of 16hz - 18khz, clearly much less than 20-20, but I think everyone can agree that the AD produce the entire spectrum in a respectable manner. This exception is probably due to the acoustic mods that Dan uses that muffs some frequencies, but the T50RP driver can produce frequencies from 15hz - 35khz according to its spec sheet so that's why the AD can easily produce all frequencies well because the actual driver itself in free space can produce sub and super sonic frequencies with ease.