Hello you audio-loving community!

In-ear-monitors seal off the surrounding noise, are extremely portable and are told to reproduce very accurate, high-quality sound. One thing I cannot really understand is the frequency response of in-ear-monitors. I have seen only one in-ear-monitor that can reproduce high frequencies up to 18khz, and that was something really expensive. All other in-ear-monitor type of headphones, of which I have seen the specs, haven't had a high frequency response beyond 16 kHz. And even the praised Ultimate Ears super.fi 5pro IEM has a high frequency response not more than 16 kHz.

My question is, how can an IEM be so good-sounding and the super.fi 5pro IEM being one of the best sounding headphones, even audiophile-quality, when they can't deliver no more than 16 kHz of high frequencies into my ear? Even the stock iPod earbuds are given a frequency response of up to 20khz.

I really confuses me because AFAIK 16khz-cutoff music sounds much worse than music reproduced up to 20khz.

Additionally, in case no more than 16khz can be heard, then I assume I could downsample my music to 32khz sapling rate without noticing much of a difference in sound quality? Or, in case of a lossy compression format, have a, say, 16,5 khz low-pass filter to save more bits for the lower frequency spectrum without noticing any quality loss in high frequencies?

Tnx in advance
Tanel

In music, anything above 16kHz is mostly noise. If you can't ABX lame -V5 --vbr-new all the time, chances are you can't hear the difference with or without the lowpass.

Indeed, according to Nyquist's theorem, you could encode with a 32,000 Hz sampling rate losslessly. The transform would not be lossless, though.
I would suggest perhaps 128kbps mp3, with a 16kHz lowpass.

Quite a lot of the quality of IEMs is due to their reducing of background noise. It improves the clarity of music considerably. Beyond that, many IEMs have extremely flat responses for headphones, at least until 10-15khz.

That said, a lot of them are still rooted in conjecture. A lot of people talk about how "fast" IEM are, their quality of midrange, etc.

I wonder how "flat" the response curves of non-IEM headphones are between 10kHz-20kHz anyway? Companies like Sony give figures of 6Hz-23kHz for $30 earbuds, which makes me wonder if the figures quoted are at 0dB or something else (especially at either extreme).

That is not the case.

You assume wrong. Above 16khz there is almost only noise. The critical listening human range is from 500Hz to 6khz.
Try this test and you'll see how hard it is to hear the difference between the original sample and 13khz lowpassed (and above) samples. Do an ABX test.

Many (probably most) males in the US and Europe are deaf at 16KHz. Its mostly younger people who can hear that low. Have you confirmed that you can even hear 16KHz? I'm a young man and was surprised to find out that my hearing cut out just above 17KHz. In a couple years I'll probably be deaf at 16k too.

Plus, even if one is able to hear an isolated tone above 16kHz, all bets are off when listening to music. I can hear tones up to 17.7kHz ( ~ 18kHz at high volume), but I'll be damned if I can detect a 16kHz lowpass on actual music.

Hello guys!
Tnx for your replies And the very interesting listening tests link! I have been there before but not thoroughly and had forgotten about it for quite some time.

Gee, I seem to have quite a special hearing. I am 18 years old so the high frequency barrier of my hearing may drop in the future, I'm just hoping it won't. I have tested cutting above 16k and noticing the difference before too. Now, knowing that we all have almost the same conditions listening to the same samples, even with CHEAP audio gear that I can currently use I can hear a clear (and annoying) difference between the original and the 16 kHz cutoff sample. 17 kHz cutoff is currently transparent for me, but I must test this again when I have more expensive audio gear to use.

Those of you who can't hear any difference from original in music quality with 12 or 13 kHz cutoff, as seems to be quite common, you just don't know what you're missing! Those samples sound so terribly bad for me. Or course, a brickwall filter sounds badly resonant, a smoother filter doesn't have that annoying effect.

PS. I'm wondering how resampling to 48k has an effect on the quality and objectivenss of evaluating the samples? I'm having sound card software issues and can't currently test it on our PRO audio soundcard (without resampling).

I have quite some more questions to ask but being sleepy, will do that in the morning.

EDIT: Though the volume was below average for me, turning the volume down a bit, 16khz becomes transparent for me too.
FYI, I was using headphones

Cheers
Tanel

Hwo loud are you listening to music?
Did you do level matched ABX tests?

If you can hear those high frequencies at a 'normal' listening level, which is very unlikely, it could be due to your gear. Let someone else make the same test at same conditions.

If you mean upsampling to 48 kHz, it can't improve the quality.

Since the headphones are currently plugged into Line Out of the integrated soundcard (it has no headphone output), the volume is below average even at full level.


Well, the levels of those mustang test samples are all equal.


That thought jumped into my mind too


I meant that maybe it in some way amplifies the resonance effect. A test sweep tone sounds very weird with native cheap sound card resampling. But I now remember I have heard the similar brickwall filter resonance effect even without resampling.


Sorry, I'm new here and while adding replies, missed your post.

Anyway, I am 18 years old. Does it mean that most 18-year-old (men) can all distinguish between a 16 kHz cutoff and the the original? And that the high frequency barrier can very quickly be very much reduced?

On the mustang samples page a hifi-oriented around-50-yo man could also hear the 16k cutoff.

I think you're right, a lot of the supposed quality of IEMs is rooted in conjecture. They might isolate from background noise in the room, but I hear my breathing and noise from the seal of the heapdhones like crazy when I wear IEMs. I mean if you take into account hearing your own breathing and the seal noises the s/n ratio with those things has got to be like 20 or 30 db. To me, for my subjective taste, that's a massive fidelity-killer right there. Put your fingers in your ears right now and see what it sounds like. Crud, that's what. That's what IEMs do. People think that's hifi? For real? Can you imagine putting on a pair of over the ear headphones and having all that noise injected into the signal and considering it hi-fi? I'm tellin' you, I just don't get it. And if you walk around the noise is like you're the jolly green giant or something. And then there's the sore ears and the difficulty taking them on and off and the difficulty of hearing what's around you. And the dirty crud from your ears. And the occasional dude who gets an infection in his outer ear. Sakes alive. But hey, whatever. I guess they do eliminate noise from the room and measure pretty flat. Enjoy.

So, I guess what I'm saying is, I wouldn't worry about whether they go over 16 khz or not.

People's opinions of IEMs differ so just because you think they sound terrible does not reflect the fact that they are actually terrible.

I have custom molds for my UM-2s and they make such a difference. I don't get any of that feeling like I'm sticking my fingers in my ears but know exactly what you mean about that feeling! I think the UM-2s cut off at 18KHz so think I'll try some listening tests too I think I'll probably deaf though....too many loud metal gigs.

These numbers mean absolutely nothing at all. Look at the frequency response of the Senheiser HD600 headphones, for example. According to Senheiser, the frequency range at -3 dB is 16 Hz to 30,000 Hz. Actual measurments show that it goes from 40 Hz to 500 Hz ! Everything else is below -3 dB.



The Mustang samples on FF123's page are currently smoothly filtered. But some time ago, they were brickwall filtered. They were replaced about one year ago, I would say.



It can create audible artifacts from inaudible frequencies (that's what the Udial sample does).