This comment is uncalled for, talk about bias. This thread is headed for a typical trainwreck.
Scott: I was about to post the same. Maybe that should have been the goal, to prove that all listening threads collapse to slanging matches between parties and get closed by the mods!
Milkshake: I still do not see how did you vote?
And yes, I had been wrong about Pavel, I ask for his pardon. I can get easily carried away, and this does not help in trying to get through the message.
But he admittedly had been playing tricks, although not dirty. This thread was full of a long discussion for an extended time of how nice a way it is to trick the partecipants in these tests.
What I cannot digest is that after he had insisted on comparisons to the original. That was properly Not Nice.
And I really think his choice of the test material was not intentional, only unfortunate.
Ciao, George
And yes, I had been wrong about Pavel, I ask for his pardon. I can get easily carried away, and this does not help in trying to get through the message.
But he admittedly had been playing tricks, although not dirty. This thread was full of a long discussion for an extended time of how nice a way it is to trick the partecipants in these tests.
What I cannot digest is that after he had insisted on comparisons to the original. That was properly Not Nice.
And I really think his choice of the test material was not intentional, only unfortunate.
Ciao, George
Absolutely no disagreement about ABX being about differentiation. But a second, interesting question is whether ones answers are lumped towards positive or negative correlation. Both point towards differentiation.
I have done a similar test with a steep 15kHz roll-off FM multiplex filter (80dB/oct IIRC). Almost all in our group correctly identified A or B. When asked for a preference we almost all preferred the rolled-off version.
But these are basically two different tests, and unless you design the test such that the frequency roll-off was the ONLY difference, you cannot draw any conclusions from that. For instance, the filter created a huge phase shift. Did we hear that? Don't know.
The important thing for us was that we could reliably differentiate between the two versions.
Jan
But these are basically two different tests, and unless you design the test such that the frequency roll-off was the ONLY difference, you cannot draw any conclusions from that. For instance, the filter created a huge phase shift. Did we hear that? Don't know.
The important thing for us was that we could reliably differentiate between the two versions.
Jan
I'm only bolding to communicate that we're totally agreeing. I didn't try to understand why I got the results I did other than I seemed to be able to hear the difference, albeit consistently flipped. 🙂
Leventhal published in 1994 an article in the JAES, where he gives a short introduction why thinking about "poor perfomance" could help to gain further insight:
Les Leventhal; Statistically Significant Poor Performance in Listening Tests, JAES Volume 42 Issue 7/8 pp. 585-587; July 1994
The reasoning is quite straight forward, if the null hypthesis postulates p=0.5 then any departure from that should ring some bells.....
Les Leventhal; Statistically Significant Poor Performance in Listening Tests, JAES Volume 42 Issue 7/8 pp. 585-587; July 1994
The reasoning is quite straight forward, if the null hypthesis postulates p=0.5 then any departure from that should ring some bells.....
In the 1970s Pioneer developed a loudspeaker by use of a listening panel. The two most popular loudspeakers of the day were the Large Advent and the JBL Century 100. These two loudspeakers were quite different in their performance. As a result the panels picked the prototype loudspeaker that was in the middle. This was sold as the model CS-R300.
In audio show room demonstrations you could play many different loudspeakers and most listeners when presented with the CS-R300s would ask if they were broken!
Panels can tell if there is a difference, but presentation and experience will bias the result.
The TL072 won the test by a fair amount, but that does not mean it is the best measuring op-amp for a follower circuit.
There can be many explanations. For example it may be the panel has become familiar with similar sounding equipment. Or it can mean that distortion measurements do not linearly follow perceived quality. It can even mean there are factors not obvious that affected the perceptions.
Or it can mean that it was near the average and so favored as was one of the worst loudspeakers to hit the market.
In audio show room demonstrations you could play many different loudspeakers and most listeners when presented with the CS-R300s would ask if they were broken!
Panels can tell if there is a difference, but presentation and experience will bias the result.
The TL072 won the test by a fair amount, but that does not mean it is the best measuring op-amp for a follower circuit.
There can be many explanations. For example it may be the panel has become familiar with similar sounding equipment. Or it can mean that distortion measurements do not linearly follow perceived quality. It can even mean there are factors not obvious that affected the perceptions.
Or it can mean that it was near the average and so favored as was one of the worst loudspeakers to hit the market.
Ed, hard to say how much one could reasonably generalize from one old anecdotal story. But here goes anyway: If there was a problem with the panel, it could have had to do with how the panel worked. If people in the panel talked to each other and reached some consensus, the cause of the product disaster might have resided in that process. Also, in Japan it could also be there were cultural aspects to how the panel worked that could have contributed to the result.
One thing that is known now about group decisions, is that people are often swayed by the views of people that talk before them.
Also, in some organizations, the boss talks first. People are often wary of openly challenging the boss, and that can interfere with what is allegedly supposed to be an objective process.
One technique to help compensate for the above-mentioned effects is to have everybody write down on a piece of paper their opinions, prior to anyone speaking. After the initial discussions, the written opinions are reviewed by the group together. At least that way, some potentially important views might not get suppressed due to the way the initial discussion may go. Especially, this type of thing may be helpful if the boss thanks people for being diligent in providing well thought through written comments. Then people can hopefully know it will be safe to give honest opinions on the next occasion.
One thing that is known now about group decisions, is that people are often swayed by the views of people that talk before them.
Also, in some organizations, the boss talks first. People are often wary of openly challenging the boss, and that can interfere with what is allegedly supposed to be an objective process.
One technique to help compensate for the above-mentioned effects is to have everybody write down on a piece of paper their opinions, prior to anyone speaking. After the initial discussions, the written opinions are reviewed by the group together. At least that way, some potentially important views might not get suppressed due to the way the initial discussion may go. Especially, this type of thing may be helpful if the boss thanks people for being diligent in providing well thought through written comments. Then people can hopefully know it will be safe to give honest opinions on the next occasion.
Please come and have a listen.
Something a little different 🙂
All is explained here:
http://www.diyaudio.com/forums/everything-else/310017-abx-testing-please-listen.html#post5131934
Something a little different 🙂
All is explained here:
http://www.diyaudio.com/forums/everything-else/310017-abx-testing-please-listen.html#post5131934
Thank you for your new test, Karl. I have posted my ABX result (12/12 correct, 0% guessing probability) into your new thread. I can achieve this result repeatedly. ABX is a great method, but just for something. In your new test, ABX brings absolutely correct result in every trial. That means, there is an audible and easily distinguishable sound difference. I am not saying what kind of difference it is.
Last edited:
All went deadly silent on all these threads, and so before I delete all my Dropbox files I'll post the image of what this file was. The first half was ww followed by tt, so Pavels original source file as I recall, and then the 1458.
It all seems so long ago now 🙂
Could any of you tell by listening ?
Attachments
thanks for great "recommendation" ...
Thanks PMA and others who had good things to say about the Cambridge Audio DacMagic Plus. A month ago I purchased one and haven't regretted it.
I finally got around to connecting mine to my old tank - Sony CDP-x779ES CD player.
The resulting sound is wonderful.
Reminds me of why we enjoy doing this stuff.
😀
mlloyd1
Thanks PMA and others who had good things to say about the Cambridge Audio DacMagic Plus. A month ago I purchased one and haven't regretted it.
I finally got around to connecting mine to my old tank - Sony CDP-x779ES CD player.
The resulting sound is wonderful.
Reminds me of why we enjoy doing this stuff.
😀
mlloyd1
Thank you mlloyd1 for your feedback. I believe that sharing experience is one of the most interesting and most important points in our hobby.
- Status
- Not open for further replies.