http://mixonline.com/recording/mixing/audi...ling/index.html

Good read. Thank you.

Some very interesting stuff in there. Two items that I can defintely concur with:

1) The higher sampling rates do not necessarily mean better sound. The primary reason IMO that DVD-A, DTS and SACD sound better is simply a better engineered recording. DTS rarely misses with their engineering quality. On the other hand I have SACDs that sound like crap and some that are excellent.

2) Room acoustics play a HUGE part in sound playback. I used to work for a company with a large RF testing chamber that was also acoustically anechoic. We put the speakers in there one weekend (JMLAB electras). the imaging was fantastic but the bass took on a very funny unnatural sound.

I have always been guided by the following 3 principles (which I borrowed)

1) Good speakers.
2) Lots of clean power amplification.
3) Good build quality.

My big disappointment is that with sagging music sales the recording companies make me pay an arm and a leg for something that should have been done right to begin with and worse yet sometimes the "audiophile" version is garbage.

I found the end of the article, talking about the sound changing by you moving just a few centimeters very interesting. It's something which most of us know and take for granted - but which is rarely taken into account in listening tests. It's the kind of argument which points out something which is forgotten because it is too obvious. Simple and trivial argument - huge consequences. Very interesting.

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

Sadly, but not at all unexpectedly, the AA crowd will grasp at any straw they can, and retail whatever half-remembered anecdote they can, to remain in denial about ABX testing.

http://www.audioasylum.com/forums/prophead...ages/43478.html

http://db.audioasylum.com/cgi/m.mpl?forum=...ight=EBradMeyer

Very interesting article. It's nothing new, only confirming a couple of tests done here where a 'hi-definition' sample could not be ABXed against a properly dithered one.

I guess the only reason to buy a 'high-definition' media is to have a source that is mastered towards audiophiles, with full dynamic range and such, because there are no quality gains to these other formats.

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.

Indeed. I think the whole time resolution thing is a bit of a red herring anyways. Saying "frequencies above 22kHz are audible given XdB of SNR" and "44.1kHz sampling has inadequate time resolution" are equivalent statements. Seperating the concepts of SNR and bandwidth from the concept of time resolution is not possible.

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!

I stumbled upon this paper today, presented by Ken Pohlmann at tape archivist's meeting a couple of years ago. In the excerpt below, isn't he presenting much the same sort of argument that you debunked, in his mentions of interaural difference and preservation of musical transient?


http://www.clir.org/activities/details/AD-...rs-Pohlmann.pdf

Emphases mine. NB I have seen the Woszczyk 2003 preprint and he too makes the same arguments (without any new listening test data).

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.

Cheers,
David.

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. 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. To me all this study demonstrates is, yes, the majority of people can't tell and therefore don't need higher sampling rates. Personally, I think to satisfy my concerns, I'd like a study that tests many people looking for any who could discern a difference, and then further testing to see how good human hearing really is. Of course, if there's a flaw in my reasoning, please, don't hesitate in letting me know.

As well, I'd just like to point out that as much as some people seem to need to justify spending money on audio, others seem to need to justify not spending money on audio. Personally, I don't care about justifications or, even other peoples' preferences; I'm only looking for what quenches my thirsty ears.

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. 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. There's also
a website supplement to the paper

http://www.bostonaudiosociety.org/explanation.htm

and Moran himself has posted here on HA in defense of the work.




I suspect he got them mostly from Woszczyk 2003, which itself turns out to be a review, rather than original research.



Which is why I find it curious that Ken Pohlmann, of all people, would be retailing these arguments. EVen more curious is the schizophrenic nature of the paper, which offers these dubious arguments up front, but devotes a later section to emphasizing why double blind listening tests are necessary to confirm
differences.

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 "JAES", one has no right to comment on it in this thread?


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

Yes, we have. It's pretty well established theory, and well recognised and used in many fields (such as radar signal processing).

From http://www.bostonaudiosociety.org/explanation.htm:

This is an interesting result. In fact, that whole page is well worth a read.

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?

Hehe.

The main flaw in 2tecs thinking is that because he isn't that experienced, he doesn't know and understand yet, that it is impossible to prove the non-existence of something anywhere in the world. But it is possible to prove the existence of something at specific locations in the world. This does NOT mean, that therefore something must exist somewhere in the world - it just means that you cannot test it. This is because we cannot look everywhere simultaneusly - we cannot test everything everywhere at the same time. Therefore, nonexistence of something regardless of location, is impossible to prove.... but it can be estimated: Probabilities. When in theory something doesn't exist, and besides of various tests and widespead awareness about the topic, no one succeeds in proving one single existence of the effect.... then it is reasonable to "asume", that it doesn't exist until proven otherwise. We have no proof that higher samplerates are unperceivable - but we also have zero evidence that it is perceivable - therefore we can ignore the issue until it starts to matter. We have no proof that the FSM doesn't exist - but we also have zero evidence that it exists - therefore we can ignore the issue until there is evidence. Thus, the burden of proof is always on the person who makes a claim about the existence of something.

And there is more to it: If apparently it is very difficult to prove the existence of something - thus, if its proposed effects seem very difficult to notice - then it is reasonable to asume, that even if it exists, its significance is very low. But if the significance of higher samplerates for listening are very low IF they exist..... then whats the point in spending all the resources for recording, storing, reproducing them? This makes higher samplerates look even more uninteresting, because it means: Higher samplerates do not seem to be perceivable - and even if they were perceivable by someone, then it is probable that in the majority of cases they are insignificant. Bummer!

First..if you're going to attempt sarcasm, it's best not to display ignorance instead -- it's 'JAES', as I wrote, which stands for the Journal of the Audio Engineering Society.

Second, Moran posted to this HA thread about the paper which I guess you were unable to call up by searching for 'moran' , like I just did.

Third, I misremembered. In fact, there was no retesting...because no subject achieved a score where p< 0.05, unless levels were jacked up to abnormal levels. Quoted from the paper (emphasis mine):

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?)

I am saying that if we cant ABX something which implements a lowpass, then why should we be able to ABX freqs which are even higher than that lowpass?

To simplify it:

IF for example we cannot ABX the removal of 18-22khz, then why should we be able to ABX the removal of 22-48khz? AFAIK, the human hearing curve doesnt go up the higher the freqs, but instead down.

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.

That might be a little exaggerated. I think the few posters you are referring to can regularly ABX select tracks, but I suspect that most tracks even they are not able to ABX a high-quality encode. And the ones that they are able to ABX they usually report such things as pre-echo, warbling, sandpaper sounds, things like that. I don't recall any case where they report it as being "less bright" or something similar that would indicate loss of high requencies.

I don't think you can make that leap. Lots of CD content rolls off above 20kHz, while some encoders keep everything up to about 19kHz, so the loss is tiny. The loss of 22-48kHz is huge in comparison. I'm not saying it's audible - I'm saying your argument is not safe.

IIRC there was one individual who could ABX a 19kHz low pass filter. This was back in the r3mix forum days, so you won't find the post on HA.

Cheers,
David.

Don't know about that....perhaps we should both review the HA archive. My memory is that a very few savants like gurubroolz are so attuned to mp3 artifacts that they CAN often ABX them, even at high CBR or VBR with the best LAME codecs...far more routinely than the average punter. It would not be surprising if mp3 codec tweakers were blessed/cursed with this talent. It would be surprising if a typically 40-ish mp3-denouncing 'audiophile' writing for Stereophile, could truly do the same.



That's why I questioned whether it was due to the HF cut.

the fact that i can hear the highest frequency available on CDs is a bit unsettling. i listen to a wide range of music, for some of which the higher frequencies often mean nothing and for others the higher frequencies are the sugar and spice of the release, including many industrial and noise albums. the fact of the matter is that i have hundreds of CDs in flac format on my harddrive (which costs about 200$) and room to spare. why not go the extra mile here and cover all frequencies audible to even the golden ears.

Good point, agreed. However, i guess you'd agree that if we had testing-material with significant content up to 22khz, speakers capable of reproducing it, and then ABX it against a 18khz-lowpassed version... that the results then do have some significance? Its no safe proof, right... but in that case, its weight would be significant, no? And its something which is much easier to test than signals >22khz, right?


I think i remember hearing about him a few years ago already. I definatelly do not envy him.


I find that statement quite unsettling as well, partially because most playback equipment isn't even capable of reproducing the full 22khz range.


Perhaps because until today, every single one of those "golden ears" failed to ABX what they can hear "easily". There isn't even one single valid and successfull >22khz DBT - nothing, zero. It is off course not impossible that you are an exception, but i guess you can understand why such stats make people suspicious unless the person can show that placebo can be excluded.

indeed, i took the test on the wrong frequency

http://www.rhintek.com/tutorial/Frequency/index.php

Eh, even if that were 22khz, this would tell you NOTHING about your hearing-capabilities, because those are test-tones. One can even engineer sounds so that you can hear even the slightest distortion regardless of your hearing capabilities (udial). So in short, test-tones are an entirely different beast than actual music - or do you listen to test-tones on your mp3-player while on the go? I can hear test-tones up to about 18khz.... but i'd never be able to perceive content that high in actual music. So in short: the true test is actual music, not testtones.

thanks for the input, but i do listen to drone music which is often a single frequency fluctuating.

lol ... sure, and you're a master of logic. Oh, and btw, thanks, when you twist it so, you only really serve to illustrate and prove my point. The flaw with your insult, is there's no 'average' number of people who have 'met' aliens. Can you use a real example that demonstrates a real flaw in my reasoning? Now that shouldn't be so hard for someone who has got it all figured out already, eh?

I'm truly sorry, I guess it's wrong to try to discuss things around here; after all, you already seem to know everything. Oh, and ridicule is such a mature and useful tactic, isn't it?

LOL. I doubt those are equivalent to test tones.

Probably less musical though.

Cheers, Slipstreem.

Well, apparently since you think it was an attempt at sarcasm, perhaps the attempt was actually completely successful? Oh, and as for the spelling thing, ya I misspelt it, however, you, apparently, still got the point.


Sigh, sorry, but how was I to know which "moran" post was the one that 'you' were talking about? From where I'm from, if you claim something, you should be the one to provide the reference, no?

Whoops.

It seems to me that this quote illustrates my point concerning the better than average listeners, no?

Perhaps you could explain this to me, since, clearly, I don't understand how you can believe that this study 'proves' that no one can hear any better than anyone else. My take on this study seems to be that a 16 bit path in an otherwise higher bit rate process simply produces no statistically significant auditory artifacts in the final product. Did I miss something?

Perhaps you could rethink the logic of that.





Getting the point, and agreeing that the point is intelligent, aren't the same thing.



A search for 'moran' brings up exactly 5 threads, one of which is this one. The others are:

AES conference London: High Resolution perception paper about listening test
Double-blind test of SACD and DVD-A vs. Redbook 16/44 in JAES September
SACD Ripping
Tired of MPC, I just have a question about OGG

Hmm, which one would likely contain input from the author of a paper about SACD vs Redbook that was published in JAES? Gosh, that's a real head-scratcher.




One might note that's a performance still no better than chance.



Actually it appears you've imagined something, though at this point I wouldn't doubt you're missing something too. You seem to think that someone here (me?), or in the paper, is saying 'no one can hear any better than anyone else'. I predict few on HA would be stupid enough to make such an unqualified claim, and I am 100% certain I would never make such a claim.

As for the paper quote above -- you haven't yet read the whole paper, have you? just checking -- yes, it may demonstrate different native discriminative ability (which alas proved irrelevant to detection of difference between SACD and SACD-->Redbook, at normal levels; even the 'best' audio pro couldn't do it better than chance at p < 0.05), but you might note and think about factoring in the different number of trials for each group. 465 for the pros (and I can guess why there'd be more for them, can you?) vs 48, 256 and 256.

I do so as well, but this is still way more complex and organic than testtones. Not even Coils "Time Machines" is similiar to testtones, although thats probably almost as droney and pure as one can get in the genre.

I really don't have the time or inclination to separate the wheat from the chaff in this thread.

Sufficing to say: can we please keep it to adult discussion, and not a petty battle of wits? If needs be I will just remove all offending posts, whether they contain a morsel of useful input or not. It would be a shame, as it confuses responses to those points.

Interesting, that's not a genre I am familiar with. Would you mind posting a (short, even 10s) sample for me to look at? I wasn't aware people listened to test tones for pleasure

Time Machines owns! however Coil's ANS is much more test tone-y

A 10secs sample wouldn't explain much, because it isn't really music in the conventional way, but mood only... and even the word mood may imply something too complex. In short, imagine expressing aurally how it feels being "almost" in narcosis... the state where you're neither unconscious nor conscious... like a blank hypnotised stare - including that slight dizzyness.... thats how time machines sounds like. Expressed simply with rather pure soundwaves which slowly change in modulation.

Coil released a lot of their material - including AFAIK time machines - in extremely low numbers (hundreds, not thousands) on their own label, with intentionally doing no reprint. Filesharing and ebay are quite probably the only ways nowadays to get Time Machines. WP Link: http://en.wikipedia.org/wiki/Time_Machines

Yes, you did, right here:

So, were 'high-scorers' able to discern if a 16 bit process was used in an otherwise 32 bit process, or not?

BTW, (and only as a reply to your off-topic ad hominem)

a) When I want to use an example, personally I always try to provide the link rather than make all the readers guess at which page I'm referring too.
b) People don't actually need to completely read the actual paper, in order to add something intelligent, sadly however, the reverse doesn't appear to be necessarily true.
c) Yes, I was indeed being sarcastic, and on more than one occasion!

That's right. Though the maths would be a bit more complicated in this case. Saying that one people did better than chance doesn't lead to a significant result.

Here's why : there is always a significance threshold above which a result is considered as successful. It is often set at "no more than 5% of probability that it was chance".
It means that in a set of random tests, no more than one out of 20, in average, would pass the test.
In this study, 60 listeners tried. In average, in the case all answers are random, then we should have got three listeners with "no more than 5% of probability that it was chance", in average !

Since this is the standard random result, maths are a bit more complicated. A much higher score is needed in order to really show a significant individual result among many other average ones.
Since the probabilities are small, we can say that the required result would have been around 5/60 = 0.08 % instead of 5%.

But even this result should not be taken as significant, because it can be seen as a post-experiment adjustment in order to fit personal convictions.
That's why when this case occurs, the usual practice is just to have the listener pass another test in order to confirm the result, and not bother with the maths.

It's like having 1000 people flip a coin ten times and having a small percentage getting heads 8 or 9 times out of 10. This is not beyond the realm of possibility. If they get heads 8 or 9 times out of 10 a second time around then maybe one can conclude the coin is biased.

In this study, people that got high scores were re-tested and failed. If they could truly hear a difference then they would have been able to repeat their high scores.

very good description, the Coil drone stuff is more than feeling, it takes you to another place altogether depending on how much attention you're willing to give it. Time Machines was is one of the greatest albums i've heard to date, its very special and is much more than simply music.

if you're looking for a taste, i'd recommend this album http://thresholdhouse.greedbag.com/release...threshold-hous/
it is a very close cousin to time machines.

What I wrote there certainly is not even close to equivalent to writing 'no one can hear any better than anyone else', and if you can't see that, you're even more obtuse than I thought. Or you're trolling.


If you'd read the earlier thread about this, or read the paper, or even read this thread you're on now more carefully, you'd know the answer. Which is that the inherent noise floor difference became audible when high system gain was applied in the test...a condition one would expect to highlight such differences...which in this case btw, were between DSD and Redbook, not 32 and 16 bit.

Are you suggesting that any single 16-bit process inserted into an otherwise 32-bit chain should be audible to 'golden ears' at normal levels in a blind test?




Go do some reading. You know where the links are, and you know how to get the paper. I'm not here to be your special ed teacher.

30 second samples are available on Last.FM.
Time Machines
One of the few pressed CDs of theirs I don't own.

I don't suppose that there are any musicians ‘out there’ willing to record a minute or so of music with 24-bit and 16-bit sample rates, and then present the tracks for us to ABX? It might be something as simple as playing back a pre-recorded sample (like karaoke background music), and then doing a recording in the two formats.

It would appear to be better to do a live take – but there is no way that a musician(s) could do it exactly the same way twice. If anyone is interested, I'll offer to host the tracks on my server. Lemme' know - it might go a long way towards assisting in a resolution to this discussion.

Andrew D.
www.cdnav.com

Why would you want it recorded twice (which introduces another variable) instead of just recording at 24-bit and dithering that version to 16-bit?

Also, what are the chances that mere mortals will have conditions with low enough background noise to make use of 24 bits?

I have to agree with pdq that introducing the variable of a different performance would undermine the ABX process.

However this thread appears to be not just about bit depth, such as three byte words of 24 bits or two byte words of 16 bits. It is also about sample rate, i.e. 44.1KHz vs a higher sampling rate.

A criticism sometimes raised is that it is unfair to derive a 44.1KHz sample from a higher sample rate rather than record direct from an analogue source. However, using such an 'unfair' derivation method (Audacity software), I was easily able to ABX (with foobar) the sound of a triangle being struck: see 'An easier exercise!' at post #68 of Listening Tests > Results for 24bit/96KHz test, vs. 16bit/44.1KHz.

An approach to address such criticism is to record one live performance but with two independent ADCs (same make and model), one set for 44.1KHz/24bits and the other for a higher sampling rate (e.g. 96KHz/24bits). [There would a slight issue of different non-linearities in the ADC process but perhaps the potential significance of that could be assessed through preliminary test recordings with both cards set at the same sample rate.]

Another issue is how filtering is implemented upon playback for different sampling rates and I think that is a variable rather difficult to evaluate the significance of. The question is: is the difference in played back sound merely a result of minor differences in actual filter implementation [recording and playback], or is it an inescapable result attributable to the use of different sampling rates, regardless of the precise filter implementation?

I would not bother with 24-bit to 16-bit conversion through noise shaped dither, but I would leave the 44.1khz recording intact at 24 bits. That would facilitate proving that a higher than 44.1KHz sampling rate can of itself make a detectable difference.