Regarding the idea that, 'no subjective view is right.' What if 'right' isn't exactly the question being asked?
Suppose we organize a trip to a symphony hall on a rehearsal night, with permission to walk the hall and listen. We sit in the 5th row, note how much direct to reverberant sound, how wide the room sounds, how fine the detail is we hear from the instruments, etc. We do that again for the 6th, and 7th rows, from the front of the first balcony, and from wings on the stage off at the side.
We then come back on performance night with tickets for the center of the 6th row. Now the room is full of people. How is the sound different from before? We try to memorize many perceptual details about the experience.
The next day we meet to listen to some speaker and room combinations, one set of speakers per room. We have a discrimination questionnaire that asks us to rate how the listening experience compares to the live performance in terms of direct to reflected sound, perceived width of the performance space, fine details of instruments, etc.
We do this over time with several hundred people. Can we learn something about what is statistically 'right' about a room and speaker combination that makes it more 'real' sounding?
Isn't that an objective measure of subjective perception, just as we can say we think 'there is an objective threshold of audibility for a population?'
Suppose we organize a trip to a symphony hall on a rehearsal night, with permission to walk the hall and listen. We sit in the 5th row, note how much direct to reverberant sound, how wide the room sounds, how fine the detail is we hear from the instruments, etc. We do that again for the 6th, and 7th rows, from the front of the first balcony, and from wings on the stage off at the side.
We then come back on performance night with tickets for the center of the 6th row. Now the room is full of people. How is the sound different from before? We try to memorize many perceptual details about the experience.
The next day we meet to listen to some speaker and room combinations, one set of speakers per room. We have a discrimination questionnaire that asks us to rate how the listening experience compares to the live performance in terms of direct to reflected sound, perceived width of the performance space, fine details of instruments, etc.
We do this over time with several hundred people. Can we learn something about what is statistically 'right' about a room and speaker combination that makes it more 'real' sounding?
Isn't that an objective measure of subjective perception, just as we can say we think 'there is an objective threshold of audibility for a population?'
Have you been to any symphony halls? They all sound different. And recordings made in those symphony halls sound different compared to live performance in the same hall. So what exactly would be compared?
Anyhow I (and I hope everybody else) go to concerts to hear the performance, not to perform silly audibility tests.
Anyhow I (and I hope everybody else) go to concerts to hear the performance, not to perform silly audibility tests.
I used to work in that field - electro acoustic and traditional modifications to concert halls, to either improve or make them suitable for different music types. While there was always broad agreement about good and bad, amongst the musicians, conductors, and acoustic consultants (who designed some of the spaces), at the detail level there was a lot of, umm... disagreement. They all tended to agree to differ though. Or go away and design a better space...
Question is which room and speakers in this experiment is judged as most 'real' sounding to the test subjects? That's all. Its not a question if reproduction is perfectly like the symphony they attended. That wasn't the research question. The question has to be chosen before the experiment is started, and the experiment has to be designed to answer the question, not some other question.
Beyond that, it may be that there are better ways to prepare, train, and warm up test subjects to judge what they perceive as more real sounding. That might be part of a preliminary study to find out how useful a trip to the symphony, a jazz band in club, whatever can help to improve focus, awareness, discrimination, etc.
Beyond that, it may be that there are better ways to prepare, train, and warm up test subjects to judge what they perceive as more real sounding. That might be part of a preliminary study to find out how useful a trip to the symphony, a jazz band in club, whatever can help to improve focus, awareness, discrimination, etc.
Last edited:
What would be the use of a study that would deem room & speakers X as most closely matching Auburn's philharmonic hall? And you still fail to understand that everybody participating in the test gives only his/hers subjective opinion. Running statistics on those subjective opinions does not make them (or the most popular opinion) objective.
It is important to get the right relays for the application. To switch audio line level signals there are special relays that have gold plating over silver bifurcated triangular contacts. This gives four points of high pressure contact crossbar positions.
The gold flash is good for the lowest level signals. If too much current is switched it will damage the gold plating and you then have silver contacts suitable for a higher current level.
For loudspeaker level relays, these are often high pressure silver button contacts. They require a much higher current carrying capacity.
The poorest contact point is often the amplifier output terminals or the loudspeaker input terminals.
I have used Omron G6A-274P series.
https://www.mouser.fi/datasheet/2/307/en_g6a-1148492.pdf
https://www.mouser.fi/datasheet/2/307/en_g6a-1148492.pdf
Once again, regarding preference versus discrimination: Its very hard to make sense out of why some people don't seem to want to understand those are two different things.
Please consider a visual analogy. You are taking a driving test to get a driver's license. The examiner asks, is the traffic light ahead red or green? If you answer, I prefer beige, then don't get your license. You have confused color preference with color discrimination. They are two different things, get it?
Please consider a visual analogy. You are taking a driving test to get a driver's license. The examiner asks, is the traffic light ahead red or green? If you answer, I prefer beige, then don't get your license. You have confused color preference with color discrimination. They are two different things, get it?
Last edited:
Discrimination made by subjective observation is still nothing more than a preference. You subjectively determine that A is better than B (another person will determine the opposite). What relevance has someone determining by subjective observation: “my DAC sounds better than any other DAC I’ve ever heard”.
Unfortunately, only measurements can be reliably compared.
No, I don’t think that quality of audio device can be determined by a single measurement.
It takes much more along with listening as a part of evaluation.
So if I see, with my own subjective eyes, a difference between red and green (which can be proven by testing),
then it is a personal preference? Hardly.
And there are statistical variations in measurement test results, just like there are in subjective test results.
The math is the same. This is a rather basic thing to understand.
then it is a personal preference? Hardly.
And there are statistical variations in measurement test results, just like there are in subjective test results.
The math is the same. This is a rather basic thing to understand.
If you take a look at cognitive science, telling a red traffic light from a green one is a subjective observation. Why? Because every human sense is subjective. All senses are processed by parts of the brain using processes that are unobservable by conscious awareness. There is a term for believing your senses, including vision, are directly showing you objective reality. The term for that illusory belief is called, Naive Realism: https://rb.gy/ag5q
Thus, as humans we cannot separate vision and hearing as one being objective and other subjective.
Last edited:
Not exactly true. Using ABX with untrained test subjects has bias towards false negatives and lower sensitivity than some other protocols. Doesn't matter if its being used for listening tests, taste tests, or whatever. However that bias can be removed with sufficient training.
Also there are other blind test protocols that are just as good or better than ABX. A/B is more sensitive to true positives, and takes little training.
Any of the protocols can be used under DBT conditions.
Who picked ABX in the first place for audio, and why? Did they know what sensory scientists know now about different protocols?
EDIT: There are also the observations raised by PMA. Fatigue and concentration. For good statistics someone may ask you to do 50 or 100 trials in row. PMA was already starting to fade after 5. Its hard, extremely hard to keep it up for a lot of trials. Small differences start to confuse and scramble the brain when things are compared over and over and over again.
Also there are other blind test protocols that are just as good or better than ABX. A/B is more sensitive to true positives, and takes little training.
Any of the protocols can be used under DBT conditions.
Who picked ABX in the first place for audio, and why? Did they know what sensory scientists know now about different protocols?
EDIT: There are also the observations raised by PMA. Fatigue and concentration. For good statistics someone may ask you to do 50 or 100 trials in row. PMA was already starting to fade after 5. Its hard, extremely hard to keep it up for a lot of trials. Small differences start to confuse and scramble the brain when things are compared over and over and over again.
From Wikipedia, the free encyclopedia
"The history of ABX testing and naming dates back to 1950 in a paper published by two Bell Labs researchers, W. A. Munson and Mark B. Gardner, titled Standardizing Auditory Tests.[1]
"A subject's choices can be on merit, i.e. the subject indeed honestly tried to identify whether X seemed closer to A or B. But uninterested or tired subjects might choose randomly without even trying. If not caught, this may dilute the results of other subjects who intently took the test and subject the outcome to Simpson's paradox, resulting in false summary results. Simply looking at the outcome totals of the test (m out of n answers correct) cannot reveal occurrences of this problem.
This problem becomes more acute if the differences are small. The user may get frustrated and simply aim to finish the test by voting randomly. In this regard, forced choice tests such as ABX tend to favor negative outcomes when differences are small if proper protocols are not used to guard against this problem."
Best practices call for both the inclusion of controls and the screening of subjects."
"The history of ABX testing and naming dates back to 1950 in a paper published by two Bell Labs researchers, W. A. Munson and Mark B. Gardner, titled Standardizing Auditory Tests.[1]
"A subject's choices can be on merit, i.e. the subject indeed honestly tried to identify whether X seemed closer to A or B. But uninterested or tired subjects might choose randomly without even trying. If not caught, this may dilute the results of other subjects who intently took the test and subject the outcome to Simpson's paradox, resulting in false summary results. Simply looking at the outcome totals of the test (m out of n answers correct) cannot reveal occurrences of this problem.
This problem becomes more acute if the differences are small. The user may get frustrated and simply aim to finish the test by voting randomly. In this regard, forced choice tests such as ABX tend to favor negative outcomes when differences are small if proper protocols are not used to guard against this problem."
Best practices call for both the inclusion of controls and the screening of subjects."
If you don't like ABx test how do you suggest to prove that you hear a difference between 2 devices? Your subjective claims are no proof by themselves.
EDIT: I don't see a huge difference between ABx and A/B testing if the sequence is truly random and no visual or audible cues are allowed. And levels are matched to 0.1dB or better.
EDIT: I don't see a huge difference between ABx and A/B testing if the sequence is truly random and no visual or audible cues are allowed. And levels are matched to 0.1dB or better.
Last edited:
Isn’t it about time now to stop talking about DBT, ABX or A/B.
This discussion does not converge to consensus, like it never did in the past.
Who needs statistical evidence when being completely happy with his audio gear.
It’s all about enjoying music and not about good, better, best.
When looking for the next step, follow your ears while staying within your budget, have an open mind, even when you select something that’s beaten to dead by others, don’t let yourself being pushed in some unwanted direction.
Just my two cents
Hans
This discussion does not converge to consensus, like it never did in the past.
Who needs statistical evidence when being completely happy with his audio gear.
It’s all about enjoying music and not about good, better, best.
When looking for the next step, follow your ears while staying within your budget, have an open mind, even when you select something that’s beaten to dead by others, don’t let yourself being pushed in some unwanted direction.
Just my two cents
Hans
Sorry Hans, but the problem is that there is no reason whatsoever to discuss audible differences (or preferences) or the validity of measurements without ensuring that the audible differences are there in the first place. Hard to see how this could be done without DBT, ABX or A/B. So if you want so stop talking about DBT, ABX or A/B you should also stop talking about audible differences, preferences or validity of measurements.
That’s not my way of looking at things. When I or someone else means to hear positive differences, its a subjective opinion that may be interesting to others to add to his list.
Example is the Marcel van der Greve Tube Dac where people think to have the best of the best.
Nice to know, it could stimulate others to also build this Dac.
A measurement when excecuted in a correct way, is by definition valid, but it’s relation to audio perception might be qustionable.
Hans
Example is the Marcel van der Greve Tube Dac where people think to have the best of the best.
Nice to know, it could stimulate others to also build this Dac.
A measurement when excecuted in a correct way, is by definition valid, but it’s relation to audio perception might be qustionable.
Hans