44.1 vs 96khz music - double blind study conducted...

[metamorphosis]

For instance this I did not know of:

Had to dig this up just to stir the sauce a little:

http://www.ocf.berkeley.edu/~ashon/audio/Ultrasoni cs.htm

It's actually a very good writeup.

Good work!

I'll add up with different sources and approaches to subliminal perception, some more critical than others for perspectives on this:

http://www.sysdesign.ca/archive/berkes_ ... eption.pdf

http://serendip.brynmawr.edu/bb/neuro/n ... hita2.html

http://serendip.brynmawr.edu/bb/neuro/n ... estha.html

http://www.csicop.org/si/9204/sublimina ... ption.html

http://serendip.brynmawr.edu/bb/neuro/n ... ellew.html

http://www.umich.edu/~onebook/pages/tab ... story.html

Just to clarify, I don't think subliminal perception fits in the same area of brain-neurology as the ultrasonics stuff-
subliminal perception works on the basis of influencing people via messages that're so low in the perceptive focus that they don't register in the brain as useful data, but of course are processed anyway.
I think that's a little different from frequencies that our ear's hear (to some extent, via interference with other frequencies to an extent also) but our brain categorises as non-useful data, and, as the article suggests, makes the sound more subjectively pleasure-generating in double-blind tests.
Similar things but I think the relevance of the subliminal perception data is somewhat arbitrary re: this particular discussion.
Cheers,
M@

[hifiboom]

I've written a long reply but forgot to press submit

I know this sucks, that why I always copy longer replies via strg-c before posting. Sometimes you get a stupid server error.

1) It's not sure that the glass will break. But the glass will break with a probability that is ridiculously close to 1
2) The study basically states that the "population average" probability of a correct guess does not differ from 0.5. However this is consistent with a simplified scenario where 99.99% of the population has no abilities to discriminate 44.1 and 96 at all but there is a 0.01% who can guess correctly with very high probability.
That means that I agree with what you said; It's possible that some people may have this ability indeed.
3) In studies showing no significant results it's very important to report their statistical power (probability to detect a significant effect given their sample size, design and an assumed 'true' effect size)

However we should not forget the population the sample was taken from. Simply put, this study in fact states that the 'average' audiophile, audio engineer or student of a related field cannot tell these both formats apart. It's a very significant finding of a well conducted study. No more no less.

I think we do aggree basically on more or less all points.

I just wanted to say that stochastical measurements are not 100%.

like for example you say 1+1=2.a mathematical fact. there is nothing like mmmh maybe with a 0,0001 probability 1+1=1,9875.

stochastical science is a probability science and I do not doubt its usefulness.

I am simply critical regarding quick conclusions coming out of such a test. Or statements that the results of such a test are 100% facts.

To give a short example:
I did a study regarding the importance of website elements with 200 probands.
And it turned out that only the most obvious elements like "more amount of pictures" and "more information" had a significant influence. Less obvious elements like a "search element" had no significant influence.
But I don't doubt the usefulness of a "search element" on a web site.

And its likely that the value of this "search element" could not be properly tested through the short test period.

Same could apply to the audio test. I don't know how much time a test person had to get used to the sample sounds @44,1 and @96khz.

It may be not unliekly that you may feel better after listening to a 96khz stream over 5 hours than you would if listening on a 44,1 khz stream.

But you maybe cannot tell a difference in a 1 minúte comparision and tell the difference.

so what i basically wanted to say from scratch: lets stay a bit critical among such a test and simply not just adopt to : ah okay, it has been tested by some guys, so there is no difference.

I'm not really against or for the test. I work at 44,1khz anyways and will never buy a 96khz audio player. I enjoy 44,1khz.

[Locus M]

Just to clarify, I don't think subliminal perception fits in the same area of brain-neurology as the ultrasonics stuff-
subliminal perception works on the basis of influencing people via messages that're so low in the perceptive focus that they don't register in the brain as useful data, but of course are processed anyway.

Well in this case it is not directed at the ultra sonic question but rather this line in the report of your link ""The experimenters found that while subjects could not recognize (i.e. perceive in the common sense of the word).."

Not percieve in "common sense" of the word is what subliminal perception is about.

(Edit:) OK, you might need a little further explaining here: If the subjects was not able to identify consciously WHAT it was that made them feel "pleasent" about the sound, you could say that whatever made them feel pleasent effected them at a subliminal level.

But on topic:
Yes and no. In the ultrasonic case I took it a good deal beyond what subliminal perception is about, namely stimuli WITHIN known range of perception but BELOW level of consciousness.

In the ultrasonic case there might be analogy because we do not know the full range or functions of all brain cells (far from it). Those we can determine, we can either scan, record cell-firings from, for instance from the eyes in monkeys with presented stimuli. But in the hearing question, those cells we can determine is usually on basis on verbal report, also as stated in the Wiki:

Audiograms in humans are produced using a piece of test equipment called an audiometer, and this allows different frequencies to be presented to the subject, usually over calibrated headphones, at any specified level. The levels are, however, not absolute, but weighted with frequency relative to a standard graph known as the minimum audibility curve which is intended to represent 'normal' hearing. This is not the best threshold found for all subjects, under ideal test conditions, which is represented by around 0 Phon or the threshold of hearing on the equal-loudness contours, but is standardised in an ANSI standard to a level somewhat higher at 1kHz[1]. There are several definitions of the minimal audibility curve, defined in different international standards, and they differ significantly, giving rise to differences in audiograms according to the audiometer used. The ASA-1951 standard for example used a level of 16.5dB SPL at 1kHz whereas the later ANSI-1969/ISO-1963 standard uses 6.5dB SPL, and it is common to allow a 10dB correction for the older standard.

Thus when determining which cells fires, it will be within this audible range, like this method demonstrates:

Georg von Bekesy demonstrated physiologically that different frequencies of sound stimulated different regions of the cochlea (Wilson 2004). Based upon the findings of Bekesy, Greenwood placed four students under the age of 29 with presumably healthy cochleas in isolation chambers and introduced pure tones within the range of audible frequencies (20-20,000Hz). Upon application of each tone, he then introduced a second pure tone of the same frequency and then raised and lowered the frequency until it was sufficiently different from the original frequency to become audible (Greenwood 1961a). Subjects responded with a handheld device allowing Greenwood to record exactly at what frequency interval the two pitches were audible and distinct. Experiments were performed over the entire range of audible frequencies (Greenwood 1961a).

Case is, that if you want to know anything about the brain, you will have to correlate your observations of it with some behavior or task performed, just looking at a lot of global areas in various scanners or cells or transmittersubstances does not makes you wiser unless you can correlate them to behavior of some kind.

In the ultrasonic question, I did not mean to suggest you could hear incredible high tones, but there seems to be possibility that the brain pick up more than we know of by virtue of the above mentioned studies because they are limited to what we report we hear and not to studies of what we hear subliminally, therefore

1. There could be detections above or below known threshold and 2. they could effect us without our awereness of it, that is at a subliminal level.

That does mean I think humans are bats and there are no physical limits, only that the (averaged) known limits of hearingsrange are uncertain to some extent besides from being estimates of average themselves.

All in all a hypothesis to be further tested.

However due to the heatened debate this "leaving a chance" for the opponents, I had better been without it seems, because people obviously thought I was going against the present test results with it, which I am not. My first post was one long support in favour of evidence. The reason for doing so, However, is that I have learned, that is always better to leave some doors open instead of making totally exclusive conclusions, because everytime there is an "evident" test in my field of work, there will be a new in time, which just as evidently states the opposite...and as such science go on and is business as usual (Thanks God! Or we would be out of work..)

I would really like it to call it the day with this one, since my best point is already stated in my post scriptum to Per, that even if this test had shown difference in perception,I would not know of to which extend it would effect our joy of music anyway.

Cheers to you!

[metamorphosis]

That does mean I think humans are bats and there are no physical limits,

Hey. HEY!!!
Only Batman is a bat. And only Chuck Norris has no physical limits. The man could kill the sun. I think he tried, but got bored, and kicked ass instead...

I get your points though, but I do think subconsciously intepreting messages is a bit different from subconsciously intepreting aural pleasure-
different parts of the brain and all that, but I take your point that the general idea is basically the same, just different bits of the brain.

m@

[jupiter8]

I still don't get it. If people don't hear the difference you can link to as many studies of ultrasonic perseption as you want. People still did'nt hear the difference.





And no: 50% of them was still right is not a valid argument. The only argument that is valid for is: I don't understand statistics and i can prove it.

[metamorphosis]

I still don't get it. If people don't hear the difference you can link to as many studies of ultrasonic perseption as you want. People still did'nt hear the difference.

You missed the point. The study I pointed to showed that people could hear the difference-
they just didn't know how they were hearing a difference.
Which is different to the study this thread was built around, which asked people to distinguish between two SR's, rather than asking them if one was more pleasurable. Ie. it asked them to measure something technical objectively rather than to measure something aesthetic subjectively, which could be used to establish whether or not higher sample rates are overall better for music.

The study the thread is built around seems rather silly by comparison, because it glosses over the central question:
do higher sample rates sound better, regardless of whether someone can differentiate by listening, between two sample rates (which is a pretty stupid exercise)? As the other study showed, higher sample rates sound better, subjectively, even if people can't identify the cause objectively.
To be fair, that study really needs to be followed up with peer-reviewed stuff...

And no: 50% of them was still right is not a valid argument. The only argument that is valid for is: I don't understand statistics and i can prove it.

Agree completely.

[Locus M]

And no: 50% of them was still right is not a valid argument. The only argument that is valid for is: I don't understand statistics and i can prove it.

Well, that was not exactly the point.
I'll let Hifiboom explain you that point if I did not express myself clearly enough:

In the case of the 44,1 and 96khz thing, it would be enough to find one person, that can pick out the 96khz signal from the 44,1khz signal, to prove that there are human beings that can hear a difference. And due to this fact a 96khz recording would have a value, unimportant of if the masses can or cannot hear a difference. audiphile products are no mass products anyways.

Thus if only one or two actually heard a difference, it would not show up in the statistics. Follow Hifibooms discussion at it. I agree with him.


Do not mind, though, I support the conclusion, but think the test is kind of misleading anyway. See my post scriptum to Per.


All the best

[Locus M]

That does mean I think humans are bats and there are no physical limits,

Hey. HEY!!!
Only Batman is a bat. And only Chuck Norris has no physical limits. The man could kill the sun. I think he tried, but got bored, and kicked ass instead...

m@

and Arrrg! I meant "does not" of course!

[metamorphosis]

Found another couple of articles while searching for drivers for my infrasonic deux simsaudio unit:

http://news.bbc.co.uk/2/hi/science/nature/3087674.stm
Experiment conducted by UK scientists using subsonic tones.

http://www.borderlands.com/newstuff/research/infra.htm
The affects of infrasonics from natural phenomenon.

Random-as shit. Not quite as relevant to the debate as it's mostly subsonics rather than > 22k, but very intriguing.

[progmatist]

I dispute those studies on 2 points: 1) They're conducted based on irrelevant criteria. They train participants to listen for the same kinds of differences between CD quality and Hi-Res one would expect to hear between MP3s and CDs. That's not where the differences lie, so of course most people won't hear any difference.

2) The conclusions are convoluted. They essentially take one participant who can hear the difference 90% of the time, another participant who hears it 10% of the time, then average the 2 together to conclude both can hear the difference 50% of the time. Furthermore, their methods are much like conducting a survey testing the public's knowledge of astronomy, but only 1% of the sample are astrophysicists. By averaging the astrophysicists with everyone else, their expertise is severely diminished.

The true advantage both analog and H-Res digital have over CD quality is twofold. 1) The mix as a whole sounds more open and transparent. The reverb in particular sounds lighter, airier, and more free flowing throughout the mix. 2) Every instrument and/or effect is better able to cut through the mix, without dominating the mix. Every instrument retreats more to its own assigned position within the sound field, resulting in a less cluttered sound.

When comparing the quality of any recordings, the one thing I listen to more than anything else is the reverb characteristics. That alone speaks volumes on overall quality.

[BONES]

Where is your evidence? You are making up bullshit to justify a debunked view. As the original article is no longer available (the link doesn't work any more), you can't possibly know what, if any training went on. You are making baseless assumptions because you cannot accept your view is wrong, no matter how much or how compelling the evidence is. It's pathetic.

[xtp]

I dispute those studies . . .

If you can hear the difference then enjoy the experience, but realise that not everybody can. I have no issues with CD quality, but I do have a dislike for the sound of some re-masters when they boost the volume so much the engineer introduces distortion—I would rather use the volume control, and my amp is from the 70’s anyway and doesn’t even have a remote. I have also owned amps I don’t like—Marantz being one make I heartily dislike—but everyone is different and I am sure there are lots of Marantz fans out there. I personally think the amp and speakers have a more profound effect on the sound of the file [within reason—getting down to 8 bit might be a bit lossy] than necessarily the file itself—put it this way; I’d rather listen to an average mp3 file on my amp [and I am not a fan of mp3 compression] than I would a 10zillion kHz file on my PM7200.
Really, it is all a matter of personal taste.

[whyterabbyt]

I dispute those studies on 2 points:

Is one of them the fact that you pointlessly resurrected a 12-year-old thread?

[BertKoor]

their methods are much like conducting a survey testing the public's knowledge of astronomy, but only 1% of the sample are astrophysicists. By averaging the astrophysicists with everyone else, their expertise is severely diminished.

Disagreed. Music is not astronomy. Everybody is qualified to have an opinion on what sounds better.

The true advantage both analog and H-Res digital have over CD quality is twofold. 1) The mix as a whole sounds more open and transparent. The reverb in particular sounds lighter, airier, and more free flowing throughout the mix. 2) Every instrument and/or effect is better able to cut through the mix, without dominating the mix. Every instrument retreats more to its own assigned position within the sound field, resulting in a less cluttered sound.

Disagreed. The words you use are highly subjective. There is no scientific base for the things you mention. An audio quality measurement tool like Rightmark Audio Analyser has no room for "position", that is a term used by deluded audiophiles. Maybe it's stereo separation and phase response you mean, but are not educated enough to recognise.

When comparing the quality of any recordings, the one thing I listen to more than anything else is the reverb characteristics. That alone speaks volumes on overall quality.

Disagreed. You're saying if there is no apparent reverb then you are less able assess the quality? Would the reverb itself not matter? For instance being a natural space or coming from DSP, either a black 19" box or VST effect?

It's not hard to do a study on solely 16bit/44.1kHz vs 24bit/96kHz vs 1bit/720kHz and do it wrong. It starts with getting the right material to listen to. Make sure no sampling rate conversion has ever happened to it, no black boxes with unknown behaviour in the signal path.



Anyway, you dug up a 12-years dormant thread on a subject bound to explode. You seem to have access to some reports, but for the rest of the world the links are long dead. Wayback machine's web archive? Care to share the links? Do you have better studies that DO prove your points? Please go ahead.

In my book you're now flagged as a user with an agenda and some bones to pick.

[FapFilter]

My current USB soundcard for my laptop sadly doesn't support more than 44.1, but back then a lot of my soft synths sounded better at higher sample rates. Of course most plugs nowadays support internal oversampling, but my uneducated guess is that native 88 or 96kHz for the whole chain should be even a tiny bit better (in theory).
This goes only for audio processing though. For pure recording and playback 44k should still be perfect and higher sample rates have more of an placebo effect as long as you are not a bat or a dog.
(Uneducated guess again, though)