1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.

    Dismiss Notice

Testing audiophile claims and myths

Discussion in 'Sound Science' started by prog rock man, May 3, 2010.
771 772 773 774 775 776 777 778 779 780
782 783 784 785 786 787 788 789 790 791
  1. Killcomic
    Hear hear!
  2. old tech
    I have never suggested that M&M or the many other scientific tests have proved that no-one under any circumstances (however contrived) can hear a difference between CD and hi res when all variables apart from bit depth and sample rates are controlled. It is you that has at times implicitly and sometimes explicitly said that these tests prove that some can hear a difference, by focussing on statistically irrelevant samples. Perhaps I am not as eloquent in my writing as @gregorio as he has summed up (many times) the fallacy of your claims.

    What all those tests have done is provided credible evidence that listeners cannot distinguish the difference between 16/44 and hi res, while over some 30 years there has been none that provide credible evidence that listeners can distinguish between the two. You then go on to focus on the one individual that scored 8/10 in one trial as some sort of proof to the contrary. Using that logic, if we did a test of a fair flip of a coin and in one trial out of 554 x 10 flips one person came up with 8 heads and two tails we should assume that we should focus on that person and how he or she flipped the coin. Well that would be a waste of time as the probability of one person flipping 8 heads out of 10 on such a large trial is well within the probability of a random event.

    Then you create a strawman by stating that "many people do not find CDs sound perfect" by conflating the container with what is in it. Sure many CDs do not sound great, just as many vinyl and DSD recordings do not sound great. No media format can magically make a poor recording or mastering or a poor production generally sound good, which is why M&M and the multitude of other supporting studies controll this critical variable. The claim that CDs are perfect is true when looking at as a format, it is naive in the extreme to suggest it then should follow that every CD should sound perfectly. All we can say is that CDs will reproduce the good or the bad production perfectly and in fact, on many imperfect productions vinyl certainly can have the edge because it is not as revealing of all the flaws in the production - a bit like comparing a transparent playback stereo with one that is coloured.

    And as for appealing to Stereophile as an authority, i think it speaks volumes for where you are coming from. A subjective publication that shuns objective measurements and listening tests, hijacked by advertisers catering to their marketing psuedoscience deceptions.
    gregorio, bigshot and bfreedma like this.
  3. Phronesis
    Regarding the superstition aspects, my sense is that a large percentage of audiophiles would change their views if they did some controlled testing for themselves, but there are multiple reasons why most audiophiles won't do them - trusting their ears, money already invested in gear, hassle of setting up controlled tests, not being sure of how to set up controlled tests, laziness, money to burn, etc.

    Regarding the elitism aspect, in my neck of the woods, and on head-fi, I can't say that I've noticed more elitism than with other consumer goods and hobbies.

    Also, I think people are people are a lot more influenced by reviews, word of mouth, and brand reputation than marketing pitches by manufacturers. I see the same thing with sports cars.

    Anyway, the holidays are almost upon us, so maybe we can all try to avoid being like this guy for a while (ignore the political aspect, I'm pointing to his general cranky attitude):

    Last edited: Dec 19, 2018
  4. Killcomic
    Oh my god that video is hilarious! Never heard of Archie Bunker before. Reminds me of an old show called Kingswood Country about a bigot and his Italian son in law (also from the 70's by the looks of it).
  5. Phronesis
    I remember watching that show when I was a kid. I was too young to really get the social messages they were portraying, but the inexplicable grumpiness of the Archie Bunker character stuck with me.
  6. old tech
    I loved that show, along with Love Thy Neighbour. Totally politcally incorrect today.
    Killcomic likes this.
  7. Killcomic
    Classic Aussie comedy! It was one of the first things I saw when I arrived in Australia. Good stuff!
  8. KeithEmo
    I think there must be some sort of confusion here.

    If you read *MY* posts carefully you will see that:
    1) I *NEVER* claimed that a definite and audible difference exists specifically between CD quality and high-res files
    2) I *NEVER* claimed that the results of Meyers and Moran proved that such a difference existed
    3) I *NEVER* claimed to know AS A FACT that any test would ever show that CDs are not "testably audibly perfect"
    5) And I disputed a few specific claims that "Meyers and Moran had shown that no such differences exist"

    I personally don't know if low-pass filtering an audio sample at 20 kHz, if done properly, can be audible or not.
    I have certainly heard a few files where it seemed to - but I have never tested it fully and properly.
    I even agree with Bigshot that there will probably not be a difference that will matter to most listeners.

    I suspect that, IF it turns out that reducing properly recorded high-resolution files to CD quality is sometimes audible, the difference will be subtle, it probably won't make a significant difference to most listeners, and it will probably only turn out to be audible to some listeners, quite possibly only on certain equipment. All I've ever claimed is that there is no legitimate justification to make a general claim that it is NEVER audible... and that a little more research is justified...

    And, yes, if you're a scientist, attempting to actually determine "if some people can control how a coin lands", and one guy throws a 8/10 you DO investigate it further. You do that because, even though that result COULD be due to random chance, it COULD also be due to a real result, and no real scientists would walk away without finding out which it is.... especially when it would only take five more minutes to have him throw the coins a few more times.

    There is also an important piece of that puzzle missing from the Meyers and Moran study.... they didn't mention how many trials they ran with each test subject. This is important because it would certainly be important to know whether the guy who was correct 8/10 that time had participated in other trials where he scored closer to random or whether that was the only trial in which he participated. And, if it turned out he had participated in other trials and scored closer to average, then you would want to know if the trials were actually properly normalized. Was he able to score very highly with one style of music, or with recordings that contained a certain instrument, or only when using "headphone #3", or WAS his one high score REALLY just random good luck. A TRUE SCIENTIST is TRYING to find something interesting - and not trying to find an excuse to rule it out. And, yes, he has to be very careful not to "find something that isn;t there", which is why we have all those standards about "repeatability" and "documenting and reporting the results in detail".

    What all those tests have done is to FAIL TO PROVIDE CREDIBLE EVIDENCE THAT ANYONE *CAN* HEAR THE DIFFERENCE.... which is quite different than providing evidence that they cannot.

    This is what you seem to not understand about statistics. If you wanted to know "if a human can run a mile in under four minutes", you could test everyone on your block, everyone in your town, and probably everyone in the entire USA, and fine nobody who could do it. And, after all those tests, you might conclude that 'it couldn't be done". You would, however, BE COMPLETELY WRONG.... because you screwed up, failed to test the one guy who we now know can run a mile in 3:43, and made an overreaching generazilation based on improper analysis of the information you had.

    Probability ALWAYS delivers a result in terms of likelihoods.... Just because a result COULD HAVE BEEN due to random chance in no way proves that it WAS due to random chance. Statistics NEVER deals in certainties and never produces results in terms of absolute facts.

    Last edited: Dec 19, 2018
  9. Killcomic
    You cannot prove something doesn't exist. To expect that is a fallacy.
    Can you prove the flying spaghetti monster does not exist?
    Can you prove I'm not typing this from my secret base on Mars?
    No you cannot!
    Just because there is a possibility of something existing does not mean you should act as if it does. On the contrary, you must work under the assumption that it does not.
    Should you stop trying to prove something exists?
    Absolutely not!
    But evidence must be solid and repeatable.
    If you're putting faith on probability alone, you are making a foolish mistake.
    Last edited: Dec 20, 2018
  10. castleofargh Contributor
    it's that time of the year where we need to pull out all the philosophical razors to dismiss ideas coming from nowhere and going right back into it.
  11. Phronesis
    I hope no one is suggesting that Santa doesn’t exist? When I was a kid, I saw ABUNDANT evidence that he’s real: wife, geography, helper elves and reindeer, high-performance flying sled, etc. Let’s not get carried away with skepticism. Some things are really real.
    GearMe likes this.
  12. castleofargh Contributor
    hey don't be hard on Stereophile. I love how they achieve to sometime directly contradict their subjective BS with their own measurements and act like everybody's right anyway. that always had a solid comedic value.
    seriously. they provided us with a great deal of measurements. I'm very grateful for that contribution. the rest, I'm not personally a fan.
  13. castleofargh Contributor
    in last week's Daily show.^_^
    "Santa is the getaway drug to believing lies because they're fun."
  14. gregorio
    1. You're joking right? YOU make a FALSE assumption based on YOUR misinterpretation and then you state the paper is erroneous because it doesn't meet your FALSE assumption. There is indeed a (MAJOR) error here, it's entirely YOURS!! This is not "The KeithEmo [FALSE] Assumption" forum .... How many times??
    1a. Then why don't you?

    2. Again, you're joking? What do you mean "if Meyer and Moran were REALLY trying to determine whether most people would hear a difference"? Meyer and Moran were not "REALLY" trying to determine that, AND you know what they were actually trying to determine because only a few lines previously you quoted them ("to determine whether people could reliably determine an audible difference when a "CD loop" was inserted into the audio signal chain"). Clearly they were NOT trying to determine "whether most people could reliably determine a difference", you have deliberately changed what they were trying to determine by adding the word "most" yourself! Where on YOUR OWN list of what this thread should be about does it state that intellectual dishonesty is acceptable? And just for the record, Meyer and Moran make it perfectly clear (numerous times) that the "people" to whom they are referring are the test subjects.

    1. The test was to determine whether their sample listeners could reliably detect an audible difference, as YOU, YOURSELF have quoted! The test was NOT to identify what those differences were, just to detect ANY difference. Therefore detailed information about what the differences were is irrelevant.
    1b. They provided an addendum to the paper when it was first presented at the AES, which provides a list of all the systems and recordings/material used in the tests (copy here http://www.bostonaudiosociety.org/explanation.htm).

    2. This is NOT the "KeithEmo Impressions" thread/forum! They were NOT reproducing "typical consumer-quality content", your "impression" is FALSE! A number of the people involved in the test were serious audiophiles and the actual content used included SACD's which the audiophiles themselves asserted; most obviously demonstrated the difference between Hi-res and CD! In addition, a Chesky "Super Audio Collection & Professional Test SACD" was used, various SACDs the paper's authors chose as likely to present a audible difference and even some original, unreleased SACD material from a SACD mastering house (where some of the tests were performed). In other words, the material used was actually the EXACT OPPOSITE of your "impression"!

    Furthermore, there were a number of additional conditions deliberately designed to be particularly favourable to detecting a difference: Very low noise floor listening environments with very high quality reproduction equipment, louder than typical/recommended listening levels (-16dBFS = 85dBSPL), TDPF dither rather than noise-shaped dither (which was standard practice for many years before the test) and test subjects which included serious audiophiles, professional music engineers and a class of young recording students (with good HF hearing). Despite these particularly favourable conditions, still no one was able to detect a difference! Again, this is NOT absolute proof that not a single person on the planet might be able to hear a difference and it is NOT intended to be, but it IS quite compelling evidence ... And it's even more compelling when compared to the evidence that differences are audible (because there is no reliable evidence to support that assertion)!

    I'm sorry but as I stated, the majority of info is not published and I cannot provide links to info that does not have any links because it is unpublished. In some/many cases doing so would actually be illegal (student data protection for example), in other cases the info/tests were specifically designed for internal use only and never to be available to the public (many studio and company tests for example) and in other cases the info might be available somewhere but I don't have the links archived and a cursory search doesn't reveal them. I would suggest starting with some standards bodies, such as the ITU or ISO and I know that the EBU has done some serious (unpublished) testing because I've been involved on one or two occasions. Also some public/state companies like the NHK and BBC have archives of their published (and some unpublished) tests/studies.

    Last edited: Dec 20, 2018
  15. old tech
    Well, we can't prove Santa doesn't exist.
771 772 773 774 775 776 777 778 779 780
782 783 784 785 786 787 788 789 790 791

Share This Page