Testing something like a dac or any piece of electrical hardware would seem to involve two difficulties. One would be the test protocol itself, and the other would undetectable switching between devices.
On the other hand, a test of what humans can hear would only necessarily involve the test protocol, since silent, undetectable switching of file sources is easily possible in software.
Regarding undetectable, hint-less switching between hardware devices, some method would have to be developed and tested first. It is possible that a solution for dacs in particular could be programmed in software, but in that case some software development might be required.
On the other hand, a test of what humans can hear would only necessarily involve the test protocol, since silent, undetectable switching of file sources is easily possible in software.
Regarding undetectable, hint-less switching between hardware devices, some method would have to be developed and tested first. It is possible that a solution for dacs in particular could be programmed in software, but in that case some software development might be required.
Last edited:
I'm sure someone would object, but one thing might be to generate one massive set of trials as 4 channels with 2 live and 2 at 0 randomly and combine them at the playback device. Each trial can be bit perfect and exactly the same in every way (but A, or B, or X?).
Jakob, was your take on those references you posted that ABX has different sensitivity and statistics or is it simply not acceptable as a protocol?
Jakob, was your take on those references you posted that ABX has different sensitivity and statistics or is it simply not acceptable as a protocol?
Of course but they would be exactly the same for both DAC's, the danger here is that the DAC's have an analog output stage gain difference and even at 64 bit floating point math digital gain loses the bit exactness. If the gain difference is enough doing it digitally could involve revisiting the dither (just guessing here) but I'm sure enough doubts will surface to sink many man-hours of work.
If there is a headphone amp that both dacs drive somehow, maybe through a summing amp, a pot in there somewhere could do it. There just needs to be way to set a comfortable level where things can be heard but ear fatigue can be minimized.
Having recently been through exercise of comparing some dacs, ear fatigue can be a problem. Got to the point it was getting hard to be sure which dac was better. We called it quits for the day and came back the next morning. Then it was easy to hear the difference. Very important not to play too loud or it's over for a period of time, IMHO.
Having recently been through exercise of comparing some dacs, ear fatigue can be a problem. Got to the point it was getting hard to be sure which dac was better. We called it quits for the day and came back the next morning. Then it was easy to hear the difference. Very important not to play too loud or it's over for a period of time, IMHO.
Last edited:
Apparently my foaming at the mouth raving and hostile challenges were off putting.
Listening in the dark makes a big difference for me, that of course means you know what.
First of all, the thread topic is about "what kind of evidence do you consider as sufficient", i fail to understand why noone of you starts a new thread to discuss the realization of a controlled listening test of whatever DUTs.
Further, what question/hypothesis do you want to address?
Scott Wurcer mentioned that he thought this thread must be about DACs as the other "Jon Boncani thread" was about DACs, but that does not really help, as the lack of a clearly expressed hypothesis was one of the problems over there.
Everything is easy for the man who don´t have to do the work.....
Further, what question/hypothesis do you want to address?
Scott Wurcer mentioned that he thought this thread must be about DACs as the other "Jon Boncani thread" was about DACs, but that does not really help, as the lack of a clearly expressed hypothesis was one of the problems over there.
Everything is easy for the man who don´t have to do the work.....
So that means this is the purely intellectual section of the problem. My answers to that question are usually fairly naive like "no peeking" and then I get dozens of pages of theory and practice on perceptual testing. I don't want to become an expert in the field I expect to find the services of someone who is to help. I actually trust that your obvious interest over the years could help.
If there is a question on signal processing theory I don't link pages of impossible (for the average DIY'er) Fourier theory and signal processing math and expect them to go away and absorb it. In fact I spend my personal time trying to find as much literature that can help folks with an average grasp of algebra and first year calculus.
Anything wrong with some intellectual activity?
I consider it as crucial; just as an example, you requested an example of a positive control, i suggested one together with an argument why it is usable as a positive control, and you answered with some "ridicule" and finally with "i don´t believe it is a good one".
So you might not like (maybe rightly) the "intellectual section" but i surely don´t like the "i don´t believe section", i hope that is understandable.
It´s your right you blame my explaining abilities (that surely can be bettered), but i have reason to doubt your "naive no peeking" assertion, pretending (at least to my understanding) that would be enough (conceded that of course the technical requirements are matched, like level matching and so on) that it is all you´d need.
Our first discussion in this forum about perceptual experiments took place in 2009 when i first mentioned Oohashi´s to you. You hadn´t enough time to study it, read only the abstract and did apparently crossreading. You complained about that he authors didn´t mentioned this, hadn´t examined that and overall it was lacking .... .
The interesting part was that you were everytime in error and every time to the disadvantage of the authors - which often hints to a bias against their work/findings - but obviously the "no peeking" part was in no way sufficient.
I guess, we all (or at least most of us) have realized that the development of a meaningfull listening experiment isn´t so easy but needs some efforts and deserves imo more than "but i don´t want to belief that" . Fair or not?
That´s fine but it is still your obligation to state upfront what you´d be willing to accept.
Well, in this thread reality (an in the other DAC thread) i certainly didn´t just cite references, but gave short summaries of the content and the implications. Further i spared the members the work of tedious calculation by providing the results for needed trial numbers.
Afair you didn´t like that either (wasn´t it "spare me the word salad"?
)I don´t know if you have noticed it, but i spend my personal to find and post the informations (and its fair to note that mmerill99 did a lot of that too) that you haven´t seen neither from other members nor from the "blind test aficionados" in general that always post "do an ABX and everything is fine" ; it was mainly written explanation to avoid the mathematically loaded more formal representation for obviously the same reasons that you did mention.
You liked to call that -if i remember- verbosity? But maybe it wasn´t you, but more on Waly´s account.
But if you like to count how often you expressed in these both threads that "obviously" it was only to "discredit" "blind testing in general" what number do you get?
Last edited:
What relevance does easiness or difficulty have to a thread about "What kind of evidence do you consider as sufficient?"? One man's difficulty may be another man's ease. Besides, individual's consideration of sufficiency is subjective. You are trying to objectify the subjectivity. It's a folly.
Afair you didn´t like that either (wasn´t it "spare me the word salad"?
No that was just a few days ago after another post like the above. Wow, I always read your posts with respect and hoped something might come of them.
Oohashi´s? do you mean the brain activity in response to ultrasonics? AFAIK that study is controversial and the results are not accepted as conclusive of anything by most of the audio community. Are you keeping a dossier on me like Mr. M, maybe you can recall pointing this out to me at the time?
EDIT - If you meant different research, my apologies.
Last edited:
Looks more like FUD attempt.
As brought up on post #21 of this thread, number 4 here: http://nyarlathotep33.free.fr/hfr/TenAudioLies.pdf
What damage does ABX box really do to the sound?
Just askin'.
IME any inserted cables/switching etc alters the resolution of the system, however little.
Whether this is an impediment to any particular ABX test would depend on variables such as SPL, magnitude of difference, type of difference (THD, IMD, FR etc), program embedded electronic noise and DUT noise.
Perhaps this ABX Box signal throughput alteration may actually be of benefit in some cases.
Dan.
Whether this is an impediment to any particular ABX test would depend on variables such as SPL, magnitude of difference, type of difference (THD, IMD, FR etc), program embedded electronic noise and DUT noise.
Perhaps this ABX Box signal throughput alteration may actually be of benefit in some cases.
Dan.
- Status
- Not open for further replies.