I think you need to focus your auditory perception on certain extracts of the recordings and listen for high pitch, bass, detail on some remarkable recordings.
If you listen even for 10 seconds and have not the listeners focus on the sound quality the chances are that no one will spot the differences in dacs, there are too many constraints which distract and affect the sound more, like the recording itself could be a lot distorted.
If you listen even for 10 seconds and have not the listeners focus on the sound quality the chances are that no one will spot the differences in dacs, there are too many constraints which distract and affect the sound more, like the recording itself could be a lot distorted.
The fact that no one could POINT at the dacs , identify them, is the result and is true.
So, does it means the iio dac is as good? does the other expensive a defective unit or the iio exceptional? For sure, there are a lot of units for thousands of dollars which can be built mainstream by the electronics giants for a fraction of the price. The reason the other dac sells for 3k$ is that it cost that much at the small production volume. So...
So, does it means the iio dac is as good? does the other expensive a defective unit or the iio exceptional? For sure, there are a lot of units for thousands of dollars which can be built mainstream by the electronics giants for a fraction of the price. The reason the other dac sells for 3k$ is that it cost that much at the small production volume. So...
I´m afraid, but for the various reasons laid out in several posts in the past we can´t conclude about the reason for this result.
Basically the result just is, that the nullhypothesis could not be rejected...
If it is so easy to hear these differences, then why are the "golden ear" people so afraid of ABX testing?
Instead of trying to invent reasons why so many people can't hear a difference, all they need to do, to make their point, is find one person who can, with statistical reliability, under controlled conditions.
Don't know what a golden ear person is. Some people seem to hear differently than others, which can include sensitivity perhaps at one extreme and virtual deafness to low level distortion at the other extreme. In terms of brain processing it isn't much different than perfect pitch recognition or some other things.
Regarding alleged fear of ABX, it seems to work pretty well for bigger differences and less well for smaller differences. Why? Seems most likely that ABX software, say, could benefit from a few simple tweaks that would help with low level sensitivity.
By the way, Earl Geddes sometimes hangs around DIYAudio and he has published some good research on what people can hear, including in the way of distortion. He says that existing research probably applies to about 95% of the population. He said to study the top 5% would require developing and validating new test methods. No work has been done on it because nobody cares enough about it to pay for it, but some very limited work I have done suggests that only fairly minor adjustments to ABX sofware might suffice. Would have to do some testing to how it works with other test subjects, though.
Unfortunately i can´t repeat every thing i´ve posted in this thread in every post, but basically there is some evidence that the ABX protocol is more prone to false negatives as other test protocols.
But, as said numerous times before, the most important thing is to be clear about the hypothesis that will be tested.
And then choose the test setting that will be the best for this purpose.
Usually there is no need to use an ABX test.
No need to invent "reasons" as so much are already known in the field of perceptual evaluation, but unfortunately a lot of people think they shouldn´t care about what is already known but should repeat every single error.
Designing good experiments, that are valid, objective and reliable is a surprisingly complex task.
I am going to make a wild guess that this is because smaller differences are harder to reliably hear when other clues have been removed. A few people may still hear a difference, but most will not even though they are sure they can hear the difference quite reliably when others clues are present.Markw4 said:
Isn't it also possible that the perceived differences result from expectation bias, and don't really exist?
The mental experience of hearing differences could easily be real. For example, suppose there are two identical files of a rock song. If for the 1st file you focus attention on the bass guitar in great detail, and notice every little fret clack, pick attack, etc. And then for the second file you focus attention on the details of the high hat as the sticks play in different locations on the top cymbal, and with different parts of the stick, the tip, the edge, etc.
You might conclude that one file has better deeper more well defined bass, and the has the attributes related to what attention was most focused on for it. It was easy to make the mistake because knowing the file name cued the focusing of attention to whatever you had originally latched onto as seeming salient.
In other words, it doesn't require hallucination to make the mistake. It only takes a lack of awareness of what a part of your mind is doing that you have no ability to directly observe. That is, the link between file name and the tendency to focus attention operates as a System 1 process outside of the awareness of System 2 (conscious awareness).
What actually happens is probably a bit more complex than the model I have described, but hopefully it is enough to illustrate what may often be strong a influencing factor.
Last edited:
Expectation bias has no agenda or purpose. The fact that it can work both ways is simply a reflection of its random nature.
Expectation bias affects everyone. I know it affects me. Its simply a reflection of how our brains are wired.
ABX testing is not about showing that a difference can be heard by all people under all circumstances. Such a proposition would be rediculous. People may not be able to hear a difference, that really exists, for a variety of reasons. However, its also possible that they can't hear a difference because the difference reported by others isn't real.
The only way this issue can be resolved is by people who can hear a difference demonstrating that fact under controlled circumstances.
The experience described by many of a difference that is clearly audible in a sighted situation vanishing under ABX testing is exactly how expectation bias works.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.