Previous discussion about the study the article mentions:
http://www.hydrogenaudio.org/forums/index....c=57406&st=

What surprises me about that article is that so many people in the "pro" audio world still believes that hi-res audio is clearly better than standard reedbook CD audio.

About the author noting the audible difference possibly due to the superior time resolution of high-res audio, he and the paper he cites are just plain wrong, which again, puzzles me. Wrong in two ways. First, it has been determined experimentally that humans can detect ITDs (interaural time delays) up to 10 us, not 15 ms, as he notes. And second, time resolution on sampling systems is not just 1/fs, 1/44100 in case of CD, but more like 1/(fs*nº of discrete levels), or 1/(44100*65535), which is thousands of times smaller than the 10 us detectable under best circumstances. So CD is more than adequate in this sense.

And as I showed a while back, Red Book has a time resolution in the *low nanoseconds* anyway, well below the limits of audibility of time resolution.

You'd think that people who cared about such things would actually test it, or quote people who test it!

Sampling Frequency
Generally, sampling frequencies of 44.1, 48, 96, and 192 kHz are used in high-fidelity
recording. The usable audio bandwidth is one-half the sampling frequency, so higher
sampling frequencies provide a wider audio bandwidth. This is potentially useful because
musical instruments can generate content with wide bandwidths; for example, a cymbal
might have response of 90 dB SPL (sound pressure level) beyond 60 kHz, and a violin
might have content beyond 100 kHz.

Even so, the use of high sampling frequencies such as 96 and 192 kHz may seem
unnecessary. In rare cases, a person may be able to hear frequencies to 24 or 26 kHz, far
below the cutoff frequencies of 48 and 96 kHz. In most cases, high-frequency hearing
response is below 20 kHz. Thus, for steady-state tones, the higher-frequency response
may not be useful. However, a high sampling frequency provides additional benefits
beyond wide audio bandwidth. It can be argued that high sampling frequencies improve
the binaural time response, leading to improved imaging in multichannel recordings. For
example, if short pulses are applied to each ear, a 15-?S difference between the pulses
can be heard, and that time difference is shorter than the time between two samples at 48
kHz. Some people can hear a 5-?S difference, which corresponds to the time difference
between two samples at 192 kHz. In theory, a high sampling frequency might improve
spatial imaging.

Similarly, higher sampling frequencies provide improved temporal response. For
example, the sampling interval at 44.1 kHz is 22.7 ?S; at 192 kHz, it is 5.2 ?S. Musical
instruments can generate transients with rise times of less than 10 ?S. As another
example, room reverberation comprises a large number of reflections arriving at high
rates. For example, reverberation might comprise 500,000 arrivals per second; spaced
regularly, this time interval is less than 2 ?S. Human subjects are sensitive to interaural
time delays of between 2 and 10 ?S. Subjects have differentiated between a regular pulse
train and one with deviations of 0.2 ?S. Higher sampling frequencies clearly preserve
temporal response (Woszczyk 2003). In addition, higher sampling frequencies allow
greater latitude in the design of the anti-aliasing low-pass filter. For example, a lowerorder
slope may be employed, providing improved time-domain response; this is further
described below. Generally, high sampling frequencies can promote improved filter and
signal processing performance in the traditional audio (0 to 20 kHz) band. Ultimately,
because the limit of human hearing acuity is not yet known, the point of transparency of a
recording system cannot be known. In some cases, such as the conversion of monaural
speech recordings, a lower sampling frequency of 48 kHz may be used. However, for
highest audio fidelity, higher sampling frequencies of 96 or 192 kHz are recommended.

Human subjects are sensitive to interaural time delays of between 2 and 10 μS.

First off, I am not an expert, however, it seems to me that there may be a 'flaw' in one of the conclusions of this study, if they only look at averages. One conclusion was that no one could tell between the two sample rates, but they don't mention how many particular individuals scored much higher than chance. If there were such individuals, then they, and they alone, would benefit from higher sampling rates.

Quote

Human subjects are sensitive to interaural time delays of between 2 and 10 ?S.

I hope they're sensitive to longer ones too!!!

Seriously 2us, where does he get that from?

And the 0.2us later on - are there any references?

10us is the usually quoted figure. I know where that came from (though don't have the reference to hand).

Of course any sampling rate is sufficient to preservse such inter-channel differences within the Nyquist bandwidth. Quantisation can impose a limit, but not anything relevant at 16-bits 44.1kHz! We've had this discussion several times before.

btw it's always better to read the actual primary research rather than dismissing it based on someone else's summary. You can buy the paper for $20 from JAES.

If I recall Meyer and Moran's paper correctly, they did retest initial 'high-scorers', who ended up doing no better than chance in multiple tests....and Moran himself has posted here on HA in defense of the work.

Of course any sampling rate is sufficient to preservse such inter-channel differences within the Nyquist bandwidth. Quantisation can impose a limit, but not anything relevant at 16-bits 44.1kHz! We've had this discussion several times before.

One of the authors, using a short repeated section of room tone on the Hartke disc mentioned above, obtained a positive result (15/15) at a gain of only 10 dB above our standard level. This setting produced sound levels clearly higher than those at the site, as the peak levels for this small vocal/percussion ensemble would have been 111 dB SPL on the loudest part of the disc.

This makes sense if you consider that for there to be those who can't hear a difference, they must be conterbalanced by those who can hear a difference.

Quote from: 2tec on 2008-04-24 03:06:31

This makes sense if you consider that for there to be those who can't hear a difference, they must be conterbalanced by those who can hear a difference.

This is an amazing use of logic and could have important application in other areas as well.

For instance, for there to be people who have never met aliens, they must be counterbalanced by those who have met aliens?

Quote from: krabapple on 2008-04-24 03:16:35

btw it's always better to read the actual primary research rather than dismissing it based on someone else's summary. You can buy the paper for $20 from JAES.

Sure, it's better, but hardly required. As for dismissing the study, I did no such thing. I'm afraid it appears that you've completely overstated my position; perhaps you didn't read my 'comment' carefully?. Oh, and as for physically purchasing the paper, are you suggesting that unless one has purchased the study from "JES", one has no right to comment on it in this thread? 

Quote from: krabapple on 2008-04-24 03:16:35

If I recall Meyer and Moran's paper correctly, they did retest initial 'high-scorers', who ended up doing no better than chance in multiple tests....and Moran himself has posted here on HA in defense of the work.

Speaking of proof, please, can you provide a link?

The test results for the detectability of the 16/44.1 loop
on SACD/DVD-A playback were the same as chance:
49.82%. There were 554 trials and 276 correct answers.
The sole exceptions were for the condition of no signal
and high system gain, when the difference in noise floors
of the two technologies, old and new, was readily audible.

As the tests progressed, we repeatedly sorted the data
for correlations with age, sex, upper frequency hearing
limit, or experience. No such correlations have emerged.
Specifically, on music at normal levels as defined here,
audiophiles and/or working recording-studio engineers got
246 correct answers in 467 trials, for 52.7% correct. Females
got 18 in 48, for 37.5% correct. Those subjects able
to hear tones above 15 kHz got 116 in 256 trials, for 45.3%
correct; listeners aged 14–25 years old (who were, as it
turned out, the same group), also got 116 correct in 256
trials, 45.3%. The “best” listener score, achieved one
single time, was 8 for 10, still short of the desired 95%
confidence level. There were two 7/10 results. All other
trial totals were worse than 70% correct.

Furthermore, none of the more elaborate and expensive
playback systems (for which the subjects were all dedicated
amateur audiophiles, active students in a professional
recording program, and/or experienced working
professionals) revealed detectable differences on music,
again at levels as defined previously.

There is something else to take into account: To a limited degree, HA-Members have already tested ultra-high frequency perceivability en masse, simply by ABXing lossy encoders. Lossy encoders use a lowpass exactly because it is asumed that people cannot hear it - at least on normal equipment. If trained ABXers cannot hear 19-22khz, then why should they be capable of hearing stuff above 22khz? In other words, we have already tested this issue en masse, simply by ABXing between lossless and lossy. Test it yourself: just lowpass a lossless file at about 18khz... then try to ABX it on your best equipment.

So you're saying that if you can't ABX lossy from lossless, you won't be able to ABX 'standard' digital audio from 'high-res' digital audio, either? (Or is this just one aspect of 'high-res' digital audio?)

There is something else to take into account: To a limited degree, HA-Members have already tested ultra-high frequency perceivability en masse, simply by ABXing lossy encoders. Lossy encoders use a lowpass exactly because it is asumed that people cannot hear it - at least on normal equipment. If trained ABXers cannot hear 19-22khz, then why should they be capable of hearing stuff above 22khz? In other words, we have already tested this issue en masse, simply by ABXing between lossless and lossy. Test it yourself: just lowpass a lossless file at about 18khz... then try to ABX it on your best equipment.

While the mass cannot generally ABX high-quality mp3s from source -- myself included in the mass -- a very few trusted HA posters (ones tending to be involved in LAME development)  have reported the ability to ABX of the highest-quality lossy encodes on a regular basis.  Don't know whether that's because of HF cut, or some other artifact.

IF for example we cannot ABX the removal of 18-22khz, then why should we be able to ABX the removal of 22-48khz?