Considering I've got a job to do and only post when I can, I've got sometimes to bow out of a thread which takes too much of my time. I don't feel much inclined to go back in when the mood of the thread is turning so deleterious.
I don't see the shortcomings in research as a failing on your part.
I still see the points you raised about auditory statistical learning as highly tangential to the criticism of ABX and a bit weak to support preference testing over a properly done ABX (when comparing DAC at least). They are interesting per se though and the papers I found made for some interesting reading.
The point of about inattentional deafness are not in question and are valid criticisms of how some ABX is done. So are the points about the fact it is easier to answer a close question with clear expectations. I already learned that for my work in social sciences so that doesn't come as a surprise. So yes, I don't feel a need to answer post 552 with which I pretty much agree and which doesn't have much to do with statistical learning.
I've stated my position. Feel free to take the last word.
PS: thx to Markw4 for being a gentleman in all of this.
@ MMerill88,
thanks for correcting my faulty recollection of Dave Moulton´s description.
@ Earl Grey,
As the usual recommendations/procedures request music sample lengths from 5s - 20s, based on somewhat mysterious auditory memory considerations, everything above is already in the "longer term" realm.
Oohahsi et al. found a considerable time lag in the EEG measurements and therefore advocated longer sample length, which is imo in good accordance with studies examining the time course on emotinal response during listening to music.
Quite new i´d say, at least within controlled experimental conditions, as imo the first published paper was:
Koreimann et al.;Inattentional Deafness Under Dynamic Musical Conditions, Proceedings of the 7th Riennial cionference of European Cognitive Sciences of Music (ESCOM 2009), p. 246 .
Isn´t your line of reasoning already an argument for more prolonged listening?
If you do more "normal" listening, consuming music songs/parts instead of short snippets, you will not concenctrate on aspect "x" all the time, instead getting a more holistic point of view.
Not to forget about the emotional response timing mentioned above.....
It seems to be already a matter of definition; is missing something like the electrical guitar not connected to a "quality aspect" .
It is quite hard to argue that listeners could totally miss an instrument in a recording but not other qualifiers, isn´t it?
would it matter wrt to his observation?
It is not so much a discussion about the merits of controlled listening tests but more about doing _sound_ controlled listening experiments. Bad ones don´t help because the "blind" property does not ensure better/correct results. As analyzing of well documented "blind" tests shows, it is as easy to get incorrect trial results as it is with "sighted" tests. One reason is that bias effects still have a strong impact under "blind" conditions.
Therefore an experimenter has to show that a test is _objective_ _valid_ and _reliable_ otherwise one cannot draw any conclusions from test results.
Doing flawed "blind" tests because in fear of flawed "sighted" tests is a contradictio in ratio. 😉
thanks for correcting my faulty recollection of Dave Moulton´s description.
@ Earl Grey,
As the usual recommendations/procedures request music sample lengths from 5s - 20s, based on somewhat mysterious auditory memory considerations, everything above is already in the "longer term" realm.
Oohahsi et al. found a considerable time lag in the EEG measurements and therefore advocated longer sample length, which is imo in good accordance with studies examining the time course on emotinal response during listening to music.
Quite new i´d say, at least within controlled experimental conditions, as imo the first published paper was:
Koreimann et al.;Inattentional Deafness Under Dynamic Musical Conditions, Proceedings of the 7th Riennial cionference of European Cognitive Sciences of Music (ESCOM 2009), p. 246 .
Isn´t your line of reasoning already an argument for more prolonged listening?
If you do more "normal" listening, consuming music songs/parts instead of short snippets, you will not concenctrate on aspect "x" all the time, instead getting a more holistic point of view.
Not to forget about the emotional response timing mentioned above.....
It seems to be already a matter of definition; is missing something like the electrical guitar not connected to a "quality aspect" .
It is quite hard to argue that listeners could totally miss an instrument in a recording but not other qualifiers, isn´t it?
would it matter wrt to his observation?
It is not so much a discussion about the merits of controlled listening tests but more about doing _sound_ controlled listening experiments. Bad ones don´t help because the "blind" property does not ensure better/correct results. As analyzing of well documented "blind" tests shows, it is as easy to get incorrect trial results as it is with "sighted" tests. One reason is that bias effects still have a strong impact under "blind" conditions.
Therefore an experimenter has to show that a test is _objective_ _valid_ and _reliable_ otherwise one cannot draw any conclusions from test results.
Doing flawed "blind" tests because in fear of flawed "sighted" tests is a contradictio in ratio. 😉
It could be a cultural thing, but I always find DF96's posts interesting, informative and entertaining
Yes, fair enough - but I don't think that is the kind of timeline (six minutes) normally meant when people advocate for "longer term" listening tests. That paper is certainly a very interesting read, and I thanked mmerill99 for providing it.
Again fair enough - it's a familiar concept in everyday life, and critical listening experience (I know can't assess how well the bass player is syncing with the kick drum if I'm listening for swirlies on the cymbals).
Well, this and the other fairly gross and conventionally understood limits to our perception are part of the reason I'm skeptical of claims to be able to differentiate devices that are far beyond "perfect enough" by objective standards. But more fundamentally than that, there's no evidence of any perceptual effect whereby the e-guitar solo is "not so available to consciousness" because of the short duration of the test (which, see below, is what I was interested in). Anyone tasked with spotting the e-guitar solo rather than counting timpani beats would identify it instantly. If the claim is only that short term tests are better given multiple cycles, I'd have a hard time disagreeing.
Not at all. No slight to the author intended; I mentioned the type of source because it's grandiose to claim that one's ideas are "verified" by "perceptual research" and then present blog posts.
To summarize:
My interest in this thread was piqued (as per my first post in it) only by a) the suggestion there were audible artifacts that were only, or perhaps primarily* detectible via longer term** listening and b) the subsequent suggestion that this was established by research.
That would be remarkable - it might even go some way to bridging the objective / subjective divide! Imagine if I could consistently differentiate two amps if I listened for a day, but if - even trained - I couldn't differentiate them in a shorter test?
But even if a) was true, b) certainly doesn't seem to be the case. I'm disappointed.
*Hard to know what "not so available to consciousness" (my emphasis) means, precisely - I would have thought that one is consciously capable of detecting something or not.
**Again, I have been taking "longer term test" to mean more than minutes.
Fair question. But if I said my definition was six minutes, I'd probably be in trouble with my "long term" partner.
We all make mistakes & it doesn't change the core of the point you made. Some people pick up on these small issues (faulty recollection) & make that the reason to suggest that your whole point is invalid (I've seen it happen elsewhere). I believe this thread has recently turned into something similar!
I agree with all of the above & the statement I made upthread about emotional response to playback Vs the way ABX is usually used (consciously identifying a difference in specific elements of the sound - cymbals, attack, ambience, soundstage, etc) I still stand by. Using my words many pages further into the thread that I can stand by all I said with research references, some will consider me over-reaching.
Fair enough!!
It's really just a reiteration of post #7 by Abraxalito:
Last edited:
JonBocani was right about one thing.
The promotion of simplicity and proven efficiency will most definitely not be allowed.
THIS was the last straw:
It seems that the Council of Elders is soon going to take action. The Council of Elders, as you may know, consists of middle-aged and elderly engineers who design amplifiers using topologies other than Class D, designers of well-regarded speakers which use midrange drivers that might significantly exceed $10 in cost, a $100 billion solid state component manufacturer that “engineers” DAC chips that sell at a broad range of prices, a toroidal transformer maker, and others, none of whom want the existence of Truthville revealed.
I understand, too, that the Mental Food Company was pretty miffed about the “mind-created steak” crack.
Let me give you real world examples then of ABX tests done by a recording/mixing engineer from here
This guy has developed a preference for high res files using sighted long-term listening.
He has posted his ABX results & his description of how he did it. Now remember he already has an preference for high-res a she find it more emotionally involving but now he has set himself the task to "prove it" using ABX. It's well worth reading this thread if people want to inform themselves about what's involved in ABX testing
Here's what he says about how he approached this particular ABX set of trials:
Remember he has developed a preference over time using sighted listening for high res files. We can see from that thread & his descriptions how difficult it is to "prove" this preference is "correct". It's very difficult because it requires examination of the two tracks forensically trying to find examples of what is different between them which will be suitable for use in ABX tests."It took me a **lot** of training. I listened for a dozen wrong things before I settled on the aspects below."
Even when this is found in sighted listening, btw (a very flawed way of listening according to some & yet being moved here to isolate a real difference), it is very difficult to keep focus on this aspect
Look at what's being said here & we all know - ABX is a test which relies on statistical analysis to estimate the likelihood one was guessing. For the number of trials done in home trials a small number of times where 'loss of focus' occurs & your statistics will NOT be statistically significant.
Does this explain my phrase that has been getting such forensic analysis & focus on
BTW, he has other published ABX test results & in these he has to use completely different types of ways of "passing" the ABX test.
I suggest that not many people are capable of doing what he is doing!!
Last edited:
BTW, as far as I can see the only one's bringing real information/evidence/research to this thread are those who know the difficulties involved in ABX testing. It seems to me the answer to this is - well you haven't convinced me so I will continue to believe that this/my ABX is not flawed
Before patting yourself too much in the back, ponder the fact that DF96 has practically helped hundreds of people on this forum while all your contributions on this forum are in heated debates of dubious value.
Sorry, can you repeat that? I didn't hear you due to the fanfare playing in my head. 😀
BTW, is that a Freudian slip? The phrase is usually "patting oneself ON the back" whereas the Antithesis of this is "stab IN the back" have you mixed up the two 😀
Last edited:
How good are your foreign languages ?
The French translation of patting oneself on the back is "se donner une tape dans le dos". No need to invoke Freud.
The French translation of patting oneself on the back is "se donner une tape dans le dos". No need to invoke Freud.
Just by the title alone this thread is calling for trouble. And this is what i see in the fast growing number of pages. But maybe, just maybe, there are things to be learnt. Lately we have known a few threads here comparing the sound of opamps by someone well respected ( Pavel from memory) and much to my surprise (+ others) a humble TL072 was given top marks. Pavel's approach was different of course - he did not present a "truth". Maybe the OP should consider recording analog out from several dacs with a good ADC and ask here which ones are prefered, without giving any clues to which is which and leave us listen instead of proposing his "Thruth" ( or the panel's) . I don' t know. When i went through the files of the opamps it was sometimes really difficult to tell which one was preferred, and i could not swear that if some astute experimenter had put several times the same file i would not beleive i perceived differences in them. I have been fooled by my ears before, i know that, but i also know that being used to live music i have yet to hear a violin through an hifi that really sounds like the real thing. Have not listened to all Hifi systems in the world neither of course.
People have different "styles". I've been accused by someone on this forum, with an abrasive and pompous manner, of taking and not giving (usually in the form of measurements and graphs and such like, which like many, he seems to be addicted to) but it's obviously a lie so I ignore it
The Gearslutz example is a good read, thanks. A critical listener who is, after training, able to make an apparently fine distinction in standard, short, blind, ABX tests.
As I mentioned above, my interest in this thread stems from the suggestion that some types of degradation / difference are not (or are "less") available "to consciousness" in a shorter test. If your point is only that sensitivity to small differences can be trained, I don't think anyone could disagree - I know I have improved my pitch recognition, for example, and I know that I can teach people to detect excessive digital compression (and dynamic compression, for that matter). That doesn't require any quasi-physiological/psychological theories about the "availability to consciousness" of audio phenomena during shorter tests. Just "critical listening is a skill can be developed".
However if someone says they have learned to appreciate some very subtle qualities of different equipment over time, and then can't then perform in an ABX test, nothing that's been presented alleviates my skepticism.
BTW, re your earlier post... ...in my first post in this thread, I asked if you had a perceptual research reference to support the earlier comment. If you didn't intend for the later remark about verification via perceptual research to apply to that, you had only to say so at that point: no harm no foul.
As I mentioned above, my interest in this thread stems from the suggestion that some types of degradation / difference are not (or are "less") available "to consciousness" in a shorter test. If your point is only that sensitivity to small differences can be trained, I don't think anyone could disagree - I know I have improved my pitch recognition, for example, and I know that I can teach people to detect excessive digital compression (and dynamic compression, for that matter). That doesn't require any quasi-physiological/psychological theories about the "availability to consciousness" of audio phenomena during shorter tests. Just "critical listening is a skill can be developed".
However if someone says they have learned to appreciate some very subtle qualities of different equipment over time, and then can't then perform in an ABX test, nothing that's been presented alleviates my skepticism.
BTW, re your earlier post... ...in my first post in this thread, I asked if you had a perceptual research reference to support the earlier comment. If you didn't intend for the later remark about verification via perceptual research to apply to that, you had only to say so at that point: no harm no foul.
This forensic examination of the quoted phrase is taken out of context so why not go back to the context HERE
In reply to this part of Jonbon's post
I replied:
I'm explaining why Jonbon's tests fail to find results which he interprets as a fault with auditory perception & I'm showing him that it's mostly a fault with ABX testing & his approach
If you think the whole issue is training then I would suggest that jonbon's test is also fatally flawed according to your view.
It's not just training in how to hear particular differences, sighted - it's also training in how to do ABX testing which involves more than you suggest
Are we actually saying the same thing then but with different emphasis & understanding?
You seem fixated on this "the suggestion that some types of degradation / difference are not (or are "less") available "to consciousness"
I have just given you a easily sensed preference that this guy had which built up over time which he found very, very difficult to differentiate in ABX testing.
We are talking about jonbon's ABX test in this thread which shows how easy it is to do a rubbish ABX test because he isn't aware of & ignores these issues.
I would suggest that JB's ABX participants are finding that "difference are not (or are "less") available "to THEIR consciousness"
If you don't understand the relevance of this to the phrase you are fixated on, there's no more I can do.
Last edited:
I haven't really followed what Jon did, what he thinks, or what you think of what Jon thinks. 🙂
But I don't think we're going to agree any further by looking at context, no, because at least some of the context - "just like we would with the higher alcohol content wine" - reinforces the implication that there is an objective physiological effect (intoxication) that always requires quantity (time, in the analogy) to come into play.
No amount of drinking can train you to get pissed on the first sip, as the Gearslutz chap has done.
But I don't think we're going to agree any further by looking at context, no, because at least some of the context - "just like we would with the higher alcohol content wine" - reinforces the implication that there is an objective physiological effect (intoxication) that always requires quantity (time, in the analogy) to come into play.
No amount of drinking can train you to get pissed on the first sip, as the Gearslutz chap has done.
You haven't read the thread but yet are commenting on text out of context??
This is looking like so much navel gazing now from you. If you want to comment on past posts, please read the thread!!
This is looking like so much navel gazing now from you. If you want to comment on past posts, please read the thread!!