ABX testing among members over the net

[soongsc]

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.

[soongsc]

One test is here

http://www.diyaudio.com/forums/showt...63#post1343463

The first sample files are here

http://www.diyaudio.com/forums/showt...22#post1342622

[Davet]

Thank you for starting this thread. I look forward to seeing the results and hopefully participating.

[auplater]

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.

[soongsc]

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

[ravon]

The ABX test was succesfully used to show there is no audible difference between different coupling capacitors.

I fully understand the wish of some people not to have this knowledge polluting their thread about selecting the best sounding capacitor.

[SY]

The ABX test was succesfully used to show there is no audible difference between different coupling capacitors.

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.

[soongsc]

Originally posted by ravon
The ABX test was succesfully used to show there is no audible difference between different coupling capacitors.

I fully understand the wish of some people not to have this knowledge polluting their thread about selecting the best sounding capacitor.

Certain resonanced or stored energy can mask audibility of very low level detail. With the sample in the original links, I could not detect differences listening directly from the PC using headphones, but when played back on my audio system, the difference became audible.

[ravon]

The question asked was: Can you hear a difference using AB comparison AND ABX testing between two sound files which were recorded using different coupling capacitors?

Can you?

[ravon]

Originally posted by soongsc
With the sample in the original links, I could not detect differences listening directly from the PC using headphones, but when played back on my audio system, the difference became audible.

I assume you did an AB comparison. What was your ABX score?