I do not know whether there is s very specific definition standard defined anywhere in a formal document for ABX testing, so I will just use the term "ABX" until any conflict arises.
During a discussion of Tripath coupling caps, this idea came up to allow some form of ABX testing of electroncis devices to be done over the net. Here is how it can be conducted.
1. The test setup consists of at least the devices to be tested, and a computer that will record the results.
2. The person interested in starting and conducting the test (The Conductor) will tell what the test setup is, what will be changed, and have signal sources prepared.
3. Participants can post music segments of 10~30 seconds of what they think be good sources for distinguishing differences for The Conductor to use if appropriate. (normally variety of music is best)
4. The conductor will play the music through each of the DUT and record the results, then create a difference file to prove that there is a difference. The 3 files will be posted so that participants may download the files for listening and post their choice of recorded music, or identify whether the difference cannot be heard.
5. When participants provide their choices, they must also post their listening setup.
6. The Conductor can wait until the appropriate numbers of feedback is reached and then announce which files are recorded from which setup.
What this kind of ABX testing does it helps identify what may influence the results, provide a more open approach to ABX, but should not be used to judge any specific member's listening setup or capability to identify differences. Each participant is encouraged to find ways to improve their listening setup if they wish based on the differences others hear.
During a discussion of Tripath coupling caps, this idea came up to allow some form of ABX testing of electroncis devices to be done over the net. Here is how it can be conducted.
1. The test setup consists of at least the devices to be tested, and a computer that will record the results.
2. The person interested in starting and conducting the test (The Conductor) will tell what the test setup is, what will be changed, and have signal sources prepared.
3. Participants can post music segments of 10~30 seconds of what they think be good sources for distinguishing differences for The Conductor to use if appropriate. (normally variety of music is best)
4. The conductor will play the music through each of the DUT and record the results, then create a difference file to prove that there is a difference. The 3 files will be posted so that participants may download the files for listening and post their choice of recorded music, or identify whether the difference cannot be heard.
5. When participants provide their choices, they must also post their listening setup.
6. The Conductor can wait until the appropriate numbers of feedback is reached and then announce which files are recorded from which setup.
What this kind of ABX testing does it helps identify what may influence the results, provide a more open approach to ABX, but should not be used to judge any specific member's listening setup or capability to identify differences. Each participant is encouraged to find ways to improve their listening setup if they wish based on the differences others hear.
One test is here
http://www.diyaudio.com/forums/showthread.php?postid=1343463#post1343463
The first sample files are here
http://www.diyaudio.com/forums/showthread.php?postid=1342622#post1342622
http://www.diyaudio.com/forums/showthread.php?postid=1343463#post1343463
The first sample files are here
http://www.diyaudio.com/forums/showthread.php?postid=1342622#post1342622
ABX screening, etc.
This looks like a "fun thing to do", but please don't present the results as some sort of valid scientifically appropriate test. There are all sorts of "gotchas" that invalidate the loose testing methods presented, not the least of which is confounded variables and lack of criteria defining what constitutes a "difference".
At best, it might establish personal bias and/or preferences in listening.
This looks like a "fun thing to do", but please don't present the results as some sort of valid scientifically appropriate test. There are all sorts of "gotchas" that invalidate the loose testing methods presented, not the least of which is confounded variables and lack of criteria defining what constitutes a "difference".
At best, it might establish personal bias and/or preferences in listening.
I agree it's not controlled enough to draw scientific conclusions, but controlled enough to get a feeling for what's going on.
Here is where I have uploaded a section of music from "Jazz at the Pawn Shop" that I often use for listening tests. Don't know how to put a link to the file, but it's under the "output temp" folder.
http://tw.briefcase.yahoo.com/soongsc
Here is where I have uploaded a section of music from "Jazz at the Pawn Shop" that I often use for listening tests. Don't know how to put a link to the file, but it's under the "output temp" folder.
http://tw.briefcase.yahoo.com/soongsc
That is incorrect, and in fact impossible. But a common misconception, and one which has led to some remarkable audiophile lore.
NO TEST can prove a negative. The correct way to describe the negative results of any test like this (whether ABX or any other double-blind format) is that "under the conditions of this test, listeners were unable to distinguish one capacitor from another by sound alone." That does NOT mean that all capacitors are audibly indistinguishable to all listeners under all circumstances, and if the authors of the test you describe were competent, I'll bet that's how they worded their conclusion.
Sorry to go off on you, but this is one of my pet peeves. Nothing personal.
For this specific case, I only was able to spend time with AB and select my preference, and am waiting for dekkersj to process more samples for others to listen for themselves. So the test has not ended yet. Please read through the previous threads. Feel free to download and listen to the samples yourself. The difference file indicates there is a difference. Some conditions may not allow the difference to be heard.ravon said:
I just focus on getting a system good enough so that differences can be heard, and the system can continously be improved.
I already mentioned that I listened to the sound files. I'm fully aware that there may be conditions which may influence the outcome of a listening test. That's why ABX testing is so powerful, the strongest influences are effectively eliminated.soongsc said:
I try to focus on things that really make a difference. There are indications that people tend to overemphasize the relevance of coupling caps on sound quality.
This thread is about ABX testing, this thread is not about keeping science away from places where you don't want it to go.soongsc said:
Be aware that if you perform an ABX test on two supplied sound files the results may be scientifically relevant.
If you read the approach in the first post, it is obvious the purpose is to use the ABX method for each participant to see if their selections have a preference trend or not, and allow readers to see how different playback systems might effect the results. The intent is to just let people make thier own decisions based on whatever they can make of the information. No need to be upset if any individual cannot hear the difference. As a matter of fact, it's even a blessing that one can take things easy and just enjoy.ravon said:
I'd be interested in seeing what others come up with after the sound tracK I posted is processed.
Please feel free to be the Conductor of a test. I would be willing to post additional music segments if necessary.
ABX
Did someone say Arnie Krueger??? I thought not...
http://www.enjoythemusic.com/hifi2005/atkinsonkrueger.html
problem as I see it is subjective arguments apply only to those making them, and are indeed valid for that person and others who agree with them. They are not valid as a generalization as to the "sound of the DUT, drug vs. placebo, etc.." applied to a population outside of those choosing to accept the results...aka general population of listeners, drug takers, etc. ABX is an attempt to set ground rules for making positive characterizations using standardized procedures that a given group agrees upon.
If you don't believe ABX works, then the results aren't valid for your set of preferences. The argument never ends, because people are arguing 2 different things.
Did someone say Arnie Krueger??? I thought not...
http://www.enjoythemusic.com/hifi2005/atkinsonkrueger.html
problem as I see it is subjective arguments apply only to those making them, and are indeed valid for that person and others who agree with them. They are not valid as a generalization as to the "sound of the DUT, drug vs. placebo, etc.." applied to a population outside of those choosing to accept the results...aka general population of listeners, drug takers, etc. ABX is an attempt to set ground rules for making positive characterizations using standardized procedures that a given group agrees upon.
If you don't believe ABX works, then the results aren't valid for your set of preferences. The argument never ends, because people are arguing 2 different things.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.