I've created two samples - one at 192000 Hz and 24 bits per sample and the other at 44100 Hz and 16 bits per sample (dithered, of course), and I wonder if anyone is able to ABX it. (I wasn't, but then again my audio equipment is not exactly high-quality...)
The interesting thing about this sample is that it actually has musical content above 22050 Hz, in fact all the way up to 96000 Hz. If you don't believe me just down-sample the 192000 Hz file until the content in-between 22050 Hz and 96000 Hz is in audible range and you'll easily verify that it's the musical content of the "plucks" of the instrument.
Here are the two samples:
192000 Hz, 24 bits/sample (http://www.sendspace.com/file/0jei7p)
44100 Hz, 16 bits/sample (http://www.sendspace.com/file/pk20xp)
Please inform me if the files become unavailable, I'll gladly upload them again.
I'm not at a location with my "good" gear but I suspect that the answer is no.
I'm not at a location with my "good" gear but I suspect that the answer is no.
I suspect that too, but hey - you never know.
Here are example shots of what the two waveforms look like at approx. 5.357 seconds:
192000 Hz 24 bps (http://img704.imageshack.us/img704/3417/192000.png)
44100 Hz 16 bps (http://img227.imageshack.us/img227/2092/44100.png)
Some of the vertical lines are not at the exact same position in the two images, probably because the zoom level was not perfectly identical for both waveforms, but that's not important.
I've created two samples - one at 192000 Hz and 24 bits per sample and the other at 44100 Hz and 16 bits per sample (dithered, of course), and I wonder if anyone is able to ABX it. (I wasn't, but then again my audio equipment is not exactly high-quality...)
The interesting thing about this sample is that it actually has musical content above 22050 Hz, in fact all the way up to 96000 Hz. If you don't believe me just down-sample the 192000 Hz file until the content in-between 22050 Hz and 96000 Hz is in audible range and you'll easily verify that it's the musical content of the "plucks" of the instrument.
Here are the two samples:
192000 Hz, 24 bits/sample (http://www.sendspace.com/file/0jei7p)
44100 Hz, 16 bits/sample (http://www.sendspace.com/file/pk20xp)
Please inform me if the files become unavailable, I'll gladly upload them again.
How many people around here even have DACs that can properly handle 192k/24bit w/o resampling, discarding the 8 least significant bits, etc.?
How many people around here even have DACs that can properly handle 192k/24bit w/o resampling, discarding the 8 least significant bits, etc.?
A lot of people here, including myself.
How many people around here even have DACs that can properly handle 192k/24bit w/o resampling, discarding the 8 least significant bits, etc.?
A lot of people here, including myself.
Well if not on hydrogenaudio then where else? :-|
That was not meant offensively, btw.
There may be no human for which the difference is audible, that's the point I'm trying to make.
There may be no human for which the difference is audible, that's the point I'm trying to make.
Well I was only curious. People claim to be able to tell the difference, I found a good sample that actually features musical content above 22050 Hz instead of just crap, so I thought why not give it an ABX test?
I also
think that it's impossible to tell the difference (I'm actually 99% sure about it), but with all the discussion about it and some people actually getting offended when you try to explain to them that they simply
can't hear the difference (you know what I'm talking about) I decided okay - let's put it to a test. I would do it myself if I had good enough hardware and speakers, but unfortunately I don't, but nevertheless it still isn't a proper test if only one person does it.
Who knows, maybe someone actually can tell the difference? I mean here's from the Wikipedia article on ultrasound:
The upper frequency limit in humans (approximately 20 kHz) is due to limitations of the middle ear, which acts as a low-pass filter. Ultrasonic hearing can occur if ultrasound is fed directly into the skull bone and reaches the cochlea without passing through the middle ear. Carefully-designed scientific studies have been performed supporting what the authors call the hypersonic effect — that even without consciously hearing it, high-frequency sound can have a measurable effect on the mind.
No I'm not trying to say that Wikipedia is the absolutely reliable source and that we should all just believe whatever is written there, and frankly I don't know what to make of that paragraph - does that mean that humans can hear sounds above 20000 Hz after all or what?
But in any case the point of this thread is just to provide an interesting high-quality sample to try to ABX. It certainly will be interesting if someone ends up being able to tell the difference, though.
But in any case the point of this thread is just to provide an interesting high-quality sample to try to ABX. It certainly will be interesting if someone ends up being able to tell the difference, though.
Auzentech X-Fi HomeTheater HD (24p/192khz native) + Sennheiser HD650 = no difference
But in any case the point of this thread is just to provide an interesting high-quality sample to try to ABX. It certainly will be interesting if someone ends up being able to tell the difference, though.
After a few attempts (quite relaxed though, ie, not really trying that hard), I couldn't distinguish them, either with an emu-0202 and AKG-K701 headphones or with JBL LSR4326P loudspeakers. I may give it a serious try tomorrow.
Since I have tried with sine waves and can't hear anything much above 17kHz, if I can ABX them, it'll be interesting (it could be due to distortion somewhere in the chain, for instance, or something more interesting--highly unlikely, though).
If you are testing in Windows 7 (and maybe Vista), be sure that the "shared mode" sampling rate is 192khz, or else you'll never hear a difference even if you are a bat
(http://meggamusic.co.uk/shup/1271299338/110314224159-Speakers-Headphones-Properties.png)
But in any case the point of this thread is just to provide an interesting high-quality sample to try to ABX. It certainly will be interesting if someone ends up being able to tell the difference, though.
Auzentech X-Fi HomeTheater HD (24p/192khz native) + Sennheiser HD650 = no difference
But in any case the point of this thread is just to provide an interesting high-quality sample to try to ABX. It certainly will be interesting if someone ends up being able to tell the difference, though.
After a few attempts (quite relaxed though, ie, not really trying that hard), I couldn't distinguish them, either with an emu-0202 and AKG-K701 headphones or with JBL LSR4326P loudspeakers. I may give it a serious try tomorrow.
Since I have tried with sine waves and can't hear anything much above 17kHz, if I can ABX them, it'll be interesting (it could be due to distortion somewhere in the chain, for instance, or something more interesting--highly unlikely, though).
Thanks for the attempt, guys.
I have a theory on why some people do in some cases manage to hear a difference between a file sampled at 192000 Hz and the one you get when you downsample it to 44100 Hz. If you don't remove all the noise above the destination Nyquist frequency (which, in case of 44100 Hz, is 22050 Hz) prior to downsampling, then aliasing is going to occur, producing sounds in audible range which previously weren't there.
For example, take this sound:
Example @ 192000Hz (http://www.sendspace.com/file/9m478u)
Don't worry, you (probably) won't be able to
hear anything, but observe what the spectrum of that sound looks like while you play it (the only frequencies are above 22050 Hz so don't do something stupid like, I dunno, use Winamp to see what the spectrum looks like).
This is what I got after I downsampled it to 44100 Hz
without any pre-filtering at all (such as removal of frequencies above 22050 Hz):
Example @ 44100 Hz (http://www.sendspace.com/file/psqo00)
You probably won't hear much in there either unless you turn up the volume a bit, but again the point is to observe the spectrum of the new signal. There are sounds all over the audible range - there were none at all in the original signal. If you'd also like to hear what the resulting file sounds like here it is normalised:
Example @ 44100 Hz normalised (http://www.sendspace.com/file/hqbdn1)
Also I'm not entirely positive that there was no pre-filtering because I don't think I can control Adobe Audition's pre-filtering really well. I'm pretty sure it does something even if you turn it off, so the actual downsampling without pre-filtering might end up being even worse than this.
I've created two samples - one at 192000 Hz and 24 bits per sample and the other at 44100 Hz and 16 bits per sample (dithered, of course), and I wonder if anyone is able to ABX it. (I wasn't, but then again my audio equipment is not exactly high-quality...)
The interesting thing about this sample is that it actually has musical content above 22050 Hz, in fact all the way up to 96000 Hz. If you don't believe me just down-sample the 192000 Hz file until the content in-between 22050 Hz and 96000 Hz is in audible range and you'll easily verify that it's the musical content of the "plucks" of the instrument.
Here are the two samples:
192000 Hz, 24 bits/sample (http://www.sendspace.com/file/0jei7p)
44100 Hz, 16 bits/sample (http://www.sendspace.com/file/pk20xp)
Please inform me if the files become unavailable, I'll gladly upload them again.
The links appear fraudulent. I was unable to download anything but advertising.
I've created two samples - one at 192000 Hz and 24 bits per sample and the other at 44100 Hz and 16 bits per sample (dithered, of course), and I wonder if anyone is able to ABX it. (I wasn't, but then again my audio equipment is not exactly high-quality...)
The interesting thing about this sample is that it actually has musical content above 22050 Hz, in fact all the way up to 96000 Hz. If you don't believe me just down-sample the 192000 Hz file until the content in-between 22050 Hz and 96000 Hz is in audible range and you'll easily verify that it's the musical content of the "plucks" of the instrument.
Here are the two samples:
192000 Hz, 24 bits/sample (http://www.sendspace.com/file/0jei7p)
44100 Hz, 16 bits/sample (http://www.sendspace.com/file/pk20xp)
Please inform me if the files become unavailable, I'll gladly upload them again.
More junk. No way are these files at all similar.
Neither of them are at all listenable.
How many people around here even have DACs that can properly handle 192k/24bit w/o resampling, discarding the 8 least significant bits, etc.?
A lot of people here, including myself.
+1. Actually, +3 - the number of audio interfaces on hand that immediately come to mind as being 24/192 capable.
The links appear fraudulent. I was unable to download anything but advertising.
You need to fix something on your end because both work fine here.
The links appear fraudulent. I was unable to download anything but advertising.
You need to fix something on your end because both work fine here.
My role in life is *not* debugging other people's pretend-freebie, come-on, watch our ads, download sites.
The samples could have been uploaded to HA itself.
The links appear fraudulent. I was unable to download anything but advertising.
You need to fix something on your end because both work fine here.
My role in life is *not* debugging other people's pretend-freebie, come-on, watch our ads, download sites.
The samples could have been uploaded to HA itself.
Considering nobody else is reporting a problem, and I was able to download them even with Noscript running, and I see no adds, and nobody asked you to debug sendspace, I agree something is borked on your end - huffy attitude or not.
How many people around here even have DACs that can properly handle 192k/24bit w/o resampling, discarding the 8 least significant bits, etc.?
I have two capable interfaces. E-MU & ASUS. Well, truth to be told, the ASUS only under certain circumstances.
The links appear fraudulent. I was unable to download anything but advertising.
You need to fix something on your end because both work fine here.
Links work fine for me. I suggest using Mozilla Firefox (http://www.mozilla-europe.org/de/firefox/) with the "Adblock Plus" Add-on (https://addons.mozilla.org/de/firefox/addon/1865) - you´ll see much less advertising then.
I have a theory on why some people do in some cases manage to hear a difference between a file sampled at 192000 Hz and the one you get when you downsample it to 44100 Hz. If you don't remove all the noise above the destination Nyquist frequency (which, in case of 44100 Hz, is 22050 Hz) prior to downsampling, then aliasing is going to occur, producing sounds in audible range which previously weren't there.
That's the responsibility of the resampler and usually taken care of. Your mileage may vary with legacy technology as Windows XP's audio architecture. Anyway, I think non-lowpassing resamplers should be much rarer than properly low-passing resamplers. Do you know just one? Also sometimes imaging above 20kHz is accepted on purpose, to get better properties (e.g. impulse response) in the 0-20kHz band.
Regarding the audibility of ultrasonic frequencies. Ultrasonic frequencies can cause audible artifacts within the audible band due to the non-linear nature of the transporting medium, e. g. air. For perfect fidelity it is not required to capture the ultrasonic frequencies directly, though. Capturing the artifacts within the audible band (if there are any at all) is sufficient. The ear wouldn't do anything else.
Who knows, maybe someone actually can tell the difference? I mean here's from the Wikipedia article on ultrasound:
The upper frequency limit in humans (approximately 20 kHz) is due to limitations of the middle ear, which acts as a low-pass filter. Ultrasonic hearing can occur if ultrasound is fed directly into the skull bone and reaches the cochlea without passing through the middle ear. Carefully-designed scientific studies have been performed supporting what the authors call the hypersonic effect — that even without consciously hearing it, high-frequency sound can have a measurable effect on the mind.
As I recall, high levels of ultrasonic energy can cause people to become more irritable, even though they do not consciously hear anything.
Makes a great selling point for high sample rates, for people who feel that thay are not irritable enough.
As I recall, high levels of ultrasonic energy can cause people to become more irritable, even though they do not consciously hear anything.
Sure, that makes sense. But why would anyone put high levels of random high frequency noise (/energy?) in that spectrum, just because they can?
As I recall, high levels of ultrasonic energy can cause people to become more irritable, even though they do not consciously hear anything.
Sure, that makes sense. But why would anyone put high levels of random high frequency noise (/energy?) in that spectrum, just because they can?
No, I´d assume because it wouldn´t make sense to filter them out when they are already there. At least, modern hardware is capable of playing that back without problems. It´s not like ultrasonic noise would cause the amp / the speaker to blow during playback. And it´s not like ultrasonic noise would introduce severe distortions (those frequencies do not desire that much power). That may have happened 40 years ago but not today. So why bother?
How many people around here even have DACs that can properly handle 192k/24bit w/o resampling, discarding the 8 least significant bits, etc.?
I have two capable interfaces. E-MU & ASUS. Well, truth to be told, the ASUS only under certain circumstances.
The links appear fraudulent. I was unable to download anything but advertising.
You need to fix something on your end because both work fine here.
Links work fine for me. I suggest using Mozilla Firefox (http://www.mozilla-europe.org/de/firefox/) with the "Adblock Plus" Add-on (https://addons.mozilla.org/de/firefox/addon/1865) - you´ll see much less advertising then.
We're being played. I tried a thrid time and I downloaded completely differents files.
So the box score is:
First attempt download failed and all I saw were on-page ads put their by the site management. They weren't pop-ups, and it is unlikely that any plug in software would selectively remove text and graphics from a page based on their content.
Second attempt appeared to suceed, but the files were garbage.
Third attempt actually produced reasonable files.
Despite all of the unhelpful, blatantly self-righteous comments, my only mistake was being prompt. I'm not expecting thanks for debugging the downloads for the late-comers.
No, I´d assume because it wouldn´t make sense to filter them out when they are already there. At least, modern hardware is capable of playing that back without problems. It´s not like ultrasonic noise would cause the amp / the speaker to blow during playback. And it´s not like ultrasonic noise would introduce severe distortions (those frequencies do not desire that much power). That may have happened 40 years ago but not today. So why bother?
Well, speakers can be a bit fragile, but any decent speaker will filter those frequencies out anyway. Still I don't see why we shouldn't filter them out - As you say "Why bother". I assume most mastering engineers have the same opinion, especially those who have been mastering that cannot hear above 15khz... That's a problem when a sine at 16khz is hammering through my head because they didn't even bother to look for obvious problems in a spectrogram!
But it seems like these samples proves that there's absolutely no reason to use such high frequencies except for mastering purposes.
They weren't pop-ups, and it is unlikely that any plug in software would selectively remove text and graphics from a page based on their content.
Actually, such software exists, though it selectively removes text and graphics based on their name and source rather than their content.
Anyways, you were just commenting on how you couldn't get the files to work, which I think is a fair claim to make. If others can figure it out or are more familiar with these services, they may disagree, but it doesn't invalidate the fact that you had difficulty. Using the upload forum here would be superior, you are absolutely correct. As a matter of fact, it would be preferred. Certain hosting sites are blocked by the forum software because they are so horrible to use.
Who knows, maybe someone actually can tell the difference? I mean here's from the Wikipedia article on ultrasound:
The upper frequency limit in humans (approximately 20 kHz) is due to limitations of the middle ear, which acts as a low-pass filter. Ultrasonic hearing can occur if ultrasound is fed directly into the skull bone and reaches the cochlea without passing through the middle ear. Carefully-designed scientific studies have been performed supporting what the authors call the hypersonic effect — that even without consciously hearing it, high-frequency sound can have a measurable effect on the mind.
As I recall, high levels of ultrasonic energy can cause people to become more irritable, even though they do not consciously hear anything.
Makes a great selling point for high sample rates, for people who feel that thay are not irritable enough.
In this case we have what appears to be a very craftsmanlike job of downsampling. The spectra of the two files pretty much overlay each other, except for the obvious difference> 22 KHz.
Above FS/2 of the 16/44 file, there is a smooth continuation of the expected 12 dB/ocatve slope that is generally characteristic of acoustic instruments. In this case the slope starts around 10 KHz.
The recording itself does not sound good to me. I doubt that there is any part of it that was made in an acoustic space using acoustic instruments.
Anyways, you were just commenting on how you couldn't get the files to work, which I think is a fair claim to make. If others can figure it out or are more familiar with these services, they may disagree, but it doesn't invalidate the fact that you had difficulty. Using the upload forum here would be superior, you are absolutely correct. As a matter of fact, it would be preferred. Certain hosting sites are blocked by the forum software because they are so horrible to use.
My point is that I eventually downloaded the files without difficulty. The second and third attempts appear to differ only in terms of wall clock time.
There never was anything wrong with my software, hardware or procedures.
The target was moved until it could be easily obtained. ;-)
I guess the most signficant thing I've learned from this experience is that some people are running their synths with really high clock speeds, for whatever purpose. Probably, they are doing it because they can. ;-)
In this case we have what appears to be a very craftsmanlike job of downsampling.
You talking about the samples from the OP? In case you are, thanks. I did the best I could.
The recording itself does not sound good to me. I doubt that there is any part of it that was made in an acoustic space using acoustic instruments.
No, it's computer-generated music (from FL Studio, in fact). Tragically though that's probably the best (and maybe the only) way acoustic frequencies above 22 kHz can turn out digitised as cleanly as they are in this sample - with microphones or any other means of recording even if the equipment is capable of capturing frequencies of up to 96 kHz they'll probably be exposed to all kinds of interferences.
Like I said, what I especially liked about this sample is that all there is is all just musical content, all the way up to 96 kHz. Unfortunately, that basically means "not recorded but rendered".
edit: Also I would like to apologise for not uploading to the forum, I'm still sort of a noob here so I didn't even know that could be done.
Who knows, maybe someone actually can tell the difference? I mean here's from the Wikipedia article on ultrasound:
Take a look at Hypersonic effect (http://en.wikipedia.org/wiki/Hypersonic_effect). Must supporting citations come from Oohashi. This topic has been discussed many times on HA.
Who knows, maybe someone actually can tell the difference? I mean here's from the Wikipedia article on ultrasound:
Take a look at Hypersonic effect (http://en.wikipedia.org/wiki/Hypersonic_effect). Must supporting citations come from Oohashi. This topic has been discussed many times on HA.
I've read that already. I think ultrasonic effect is just a pile of horsecrap, but I've found a good sample containing HFC and wanted to see if anyone actually
can tell the difference, just in case the ultrasonic effect is not a pile of horsecrap.
If you are testing in Windows 7 (and maybe Vista), be sure that the "shared mode" sampling rate is 192khz, or else you'll never hear a difference even if you are a bat
(http://meggamusic.co.uk/shup/1271299338/110314224159-Speakers-Headphones-Properties.png)
I thought if you set it like that, ANYTHING (unless through WASAPI exclusive) you play will be resampled to 192KHz? Or am I wrong?
I thought if you set it like that, ANYTHING (unless through WASAPI exclusive) you play will be resampled to 192KHz? Or am I wrong?
I have never used Windows Vista nor Windows 7 myself, but why would that be a bad thing? The only way resampling to 192 kHz prior to reproducing the sound could damage the sound quality is if Microsoft wrote a REALLY awful resampling algorithm, which...
Hm, actually you make a good point.
Here are the two samples:
192000 Hz, 24 bits/sample (http://www.sendspace.com/file/0jei7p)
44100 Hz, 16 bits/sample (http://www.sendspace.com/file/pk20xp)
foo_abx 1.3.4 report
foobar2000 v1.0.3
2010/06/07 15:40:10
File A: C:\Documents and Settings\Administrador\Mis documentos\Downloads\Sample_192000Hz_24bit.flac
File B: C:\Documents and Settings\Administrador\Mis documentos\Downloads\Sample_44100Hz_16bit.flac
15:40:10 : Test started.
15:41:18 : 01/01 50.0%
15:41:24 : 02/02 25.0%
15:41:51 : 03/03 12.5%
15:42:02 : 04/04 6.3%
15:42:36 : 05/05 3.1%
15:42:43 : Test finished.
----------
Total: 5/5 (3.1%)
Headphones Sennheiser HD 447. Soundcard Audigy SE 24/96.
foobar -> output -> 24 bits
The difference is quite simple to spot. Sample 192/24 has much clear, sharp sound. (I undertsand that it might sound like placebo or troll but there is ABX log)
Maybe there is something wrong with my hardware.
The other question is how red book sample was downsampled.
Uhmm...did you decide in advance to only try five times, or did you quit as soon as the probability dropped below 5%?
If the latter then this is not an acceptable way to perform the test. Not saying that you can't hear the difference, only that you are not following standard procedure.
One other thing - often the apparent difference is actually caused by a deficiency in the hardware used to reproduce the samples.
If you are testing in Windows 7 (and maybe Vista), be sure that the "shared mode" sampling rate is 192khz, or else you'll never hear a difference even if you are a bat
(http://meggamusic.co.uk/shup/1271299338/110314224159-Speakers-Headphones-Properties.png)
is that really necessary when running Asio?
is that really necessary when running Asio?
No
One other thing - often the apparent difference is actually caused by a deficiency in the hardware used to reproduce the samples.
Indeed - I have witnessed this myself, and to be worse the problem was not even in my hardware but in the software and surprisingly that software was Adobe Audition 3.0.1! I have created a sample which contains only noise above 20 kHz, in fact this is the sample I'm talking about (http://www.speedyshare.com/files/22868859/download/sample.flac).
In Cool Edit Pro 2.1 when I played it back I've heard, of course - nothing. Whether my sound card or my speakers actually produced the sound I have no idea, but that is completely irrelevant because what's important is that when I played back that sample on my system - I heard nothing, regardless if it was my speaker's or my soundcard's inability to reproduce noises that high or my inability to perceive them.
However, when I loaded that very same sample in Adobe Audition 3.0.1 and played it back, it ended up sounding something like this (http://www.speedyshare.com/files/22868882/download/sample%20played%20back.flac). Unfortunately the only way for me to capture the noise that I was hearing was to let Adobe Audition play back the original sample and record the output with another application (not recording with microphone what came out of my speakers but directly recording the soundcard output), so this is not the perfect replica of the noise I was hearing but certainly close enough. It's obvious what's going on here - since Adobe Audition apparently doesn't even attempt to produce any noises above 24 kHz (the noises above that I ascribe to playback recording artefacts) which is a Nyquist frequency of 48 kHz sampling period, the inescapable conclusion is that it instead downsamples the output automatically on-the-fly to 48 kHz
but without any filtering of frequencies above 24 kHz to prevent downsample aliasing, which on playback results in this noise.
Therefore, make sure that something like this isn't happening on your system neither because of the software nor because of the hardware you're using. To test if it's happening to you, download and play back the sample I was talking about (this one (http://www.speedyshare.com/files/22868859/download/sample.flac)) on the system you were doing the ABX test on and with the same software you were doing the ABX test with and see if it sounds anything like this (http://www.speedyshare.com/files/22868882/download/sample%20played%20back.flac).
Are you sure that you had configured Audition's audio output device correctly?
Are you sure that you had configured Audition's audio output device correctly?
As a matter of fact I did try it and never succeeded to make it reproduce 192k sampling rate samples properly but now that you asked I decided to try again and with some tedious tinkering managed to force it into doing it but at the moment it seems incredibly unstable because it already fell back to the previous faulty behaviour twice and I just can't seem to figure out what makes it fall back to it. Regardless of that though it serves as an example that software faults are possible even if you have a fully capable hardware, meaning that it's possible the above ABXing is invalid.
Headphones Sennheiser HD 447. Soundcard Audigy SE 24/96.
foobar -> output -> 24 bits
The difference is quite simple to spot. Sample 192/24 has much clear, sharp sound. (I undertsand...
I understand the Audigy SE 24/96 cannot handle 192/24 (or 24 /192) samples without.. re-sampling?
If you are testing in Windows 7 (and maybe Vista), be sure that the "shared mode" sampling rate is 192khz, or else you'll never hear a difference even if you are a bat
I thought if you set it like that, ANYTHING (unless through WASAPI exclusive) you play will be resampled to 192KHz? Or am I wrong?
Use an exclusive mode media player? Else you are correct, in shared mode everything that is NOT at the selected rates, gets resampled.