I published some of my own results elsewhere (have you bothered to read my testing article?), but also have never made extraordinary claims that require it as backup.
I did participate in two of Jan's tests, both on phenomena that I found interesting, and both using other DBT formats, but with proper controls. I've posted about those as well, and you're free to search.
Any more excuses? If you can't sort Mooly's files, that's fine, it's a data point. But acting butthurt because someone else took the trouble to use appropriate controls while you wouldn't is unbecoming of you.
I did participate in two of Jan's tests, both on phenomena that I found interesting, and both using other DBT formats, but with proper controls. I've posted about those as well, and you're free to search.
Any more excuses? If you can't sort Mooly's files, that's fine, it's a data point. But acting butthurt because someone else took the trouble to use appropriate controls while you wouldn't is unbecoming of you.
Okay, have sent a full sorting to Karl now, so he's got at least one complete of results to play with ... 
On a re-listen, my appraisal of A and B didn't change, didn't bother to listen again to my earlier choices, I was happy with my approach.
What I was doing, which may be of benefit to some, was that I was running the listening at maximum volume - it was easy to hear all the extraneous noises of the musician's bodies, and the physical interaction with their instruments - and I was particularly picking on the tonal quality of the quick little runs on the piano, 5 or 6 or whatever notes, that are done regularly with the right hand, throughout the piece - this was a very strong indicator ...
On a re-listen, my appraisal of A and B didn't change, didn't bother to listen again to my earlier choices, I was happy with my approach.
What I was doing, which may be of benefit to some, was that I was running the listening at maximum volume - it was easy to hear all the extraneous noises of the musician's bodies, and the physical interaction with their instruments - and I was particularly picking on the tonal quality of the quick little runs on the piano, 5 or 6 or whatever notes, that are done regularly with the right hand, throughout the piece - this was a very strong indicator ...
Last edited:
Okay, have sent a full sorting to Karl now, so he's got at least one complete of results to play with ...
I think you don't understand the situation and your assumption is wrong. I would say this is a null test, or if not, this is unnecessarily difficult that it will take too much effort to do a proper listening and for what??
Not long before or after SY writing in the Linear Audio, I think it was Hawkford saying something like this: "Of course, it is very easy to get a null result from a DBT, and it may be a good method if that is your agenda." Hawkford has a correct understanding of the situation imo. You need HUGE time/effort if you want to equate null result with inaudibility.
Even tho I believe I can do Foobar ABX better than anyone else, I will say that this is a null test, unless may be if huge effort is put into it, but I will still think that the result will be null.
The question is why do you want to know the difference between a cheap film cap with five electrolytics in series? There is probability that the electrolytics sound better. How about comparing 5 electrolytics in series with 50 electrolytics in series?
1) The difference will be very very small, smaller than difference that is created during the process (as PMA has mentioned). And this is very true.
2) The film cap is very cheap. Why not put a Mundorf or an Auricap here? We need to prove if the cost worth anything right? And who is thinking about replacing a cap with five electrolytics or the opposite? Two electrolytics back to back is better than one. Two film caps in series may be better than one. So five electrolytics could be better than just one. So what are we comparing here? Is the answer worth the effort?
3) The music is not suitable to easily find sound difference, technically and psychologically. Psychologically it is known that the respondents should better like the music. Only a few likes to listen violin.
4) I said this is a strawman test because there is an expected conclusion (which is a null-result) and the process is tailored to achieve that conclusion (either due to ignorance or political agenda or both). Why not value others time and give them a more useful and valid conclusion? I haven't listened to proper a system after my amp blown several weeks ago. And I'm building 2 or 3 class-A amps to be finished within 4 days. Anyhow I have downloaded A,B,X,Y and listened through a cheap mono speaker, 1 Watt chip amp and a borrowed $300 laptop.
I could hear differences between X and Y (but both are horrible, with their own weaknesses), but NOT with A and B. So it is interesting that Frank could hear AB and found XY harder.
BTW, I can do a good ABX between A and B but I will not tell you why I think it is useless (Read Hawkford if you want to know, if you are able to understand).
Jay, it's all about what one listens for - if I change my 'angle' so to speak I might find X and Y more straightforward, but on the first, fairly thorough run through there weren't really obvious things, for me. With A and B I barely registered the violin, the piano told me what I wanted to know ...
Last edited:
My second attempt (7/7, 0.8% probability of guessing). The longer I listened the harder to pick up differences (fatigue I guess). I can hear different characteristics of the sound of the two files A and B, but I'm not sure I can put in the right group of one file without relative comparison with another file. I mean, when I hear file "x" I can't be sure if it belongs to group "R" or "B", but when I hear file "y" as comparator I knew.
So the next step is to listen to A and B outside of ABX module so I can put A and B into either "R" or "B" groups.
ADD:
Okay so A is in group "B" and B is in group "R". Now going to download the 12 files.
So the next step is to listen to A and B outside of ABX module so I can put A and B into either "R" or "B" groups.
ADD:
Okay so A is in group "B" and B is in group "R". Now going to download the 12 files.
Attachments
Last edited:
We will see how far the unexpected added variables will ruin the characteristics of files A and B...
For example, file A is "heavy" and file B is "light". It is easier to know that file A is "heavy" when there is file B for comparison. And now we have file C. Will file C be either as heavy as A or as light as B? what if heavier than A? What if heavier than B but lighter than A?
Assume we will prove this with ABX. We have 5/5 for differentiating AC and we have 5/5 for differentiating BC. Will this be a proof that C is not equal to either A or B, or that the test failed? Now we have difficulty because I'm sure we can successfully ABX two "identical" files, if only 5/5.
Now for these "characteristics" of file A and B, is there any known characteristics of Shizuki film cap? (or 5 electrolytics in series?). From my experience, different brands (remember ESR/ESL) can have characteristics of A or B. I'm also not familiar with the sound of cheap caps, especially 5 electros ion series. So it is difficult to say that any available characteristics is caused by DUT and not by something else.
But well, we will see later. Dan, hopefully you join this test (because you have good ears) so we can answer the question.
ADD:
BTW, I'm not as lucky as Frank or other who can listen for more "signs" in the music as I have to listen in mono and from 90 degree off axis (difficult to move my stuffs).
Last edited:
This will require some vacation time to get thru all of this, what was Mooley thinking..
They are definitely different and Mooly, that sly basturd
Who ? moi
Yes they are all different but can you match those differences. This is a simple screen shot of A and B running together. The files are near perfect in time alignment. The resolution of a picture like this I would guess misses much of the really fine stuff that you might see on a scope but it gives an idea.
Attachments
Okay, have sent a full sorting to Karl now, so he's got at least one complete of results to play with ...
On a re-listen, my appraisal of A and B didn't change, didn't bother to listen again to my earlier choices, I was happy with my approach.
What I was doing, which may be of benefit to some, was that I was running the listening at maximum volume - it was easy to hear all the extraneous noises of the musician's bodies, and the physical interaction with their instruments - and I was particularly picking on the tonal quality of the quick little runs on the piano, 5 or 6 or whatever notes, that are done regularly with the right hand, throughout the piece - this was a very strong indicator ...
That's great Frank, thanks. Your the first to do the full set
I'm going to collate your results later and see how they match.
Thanks for joining in Jay. Your ABX results are outstanding but what you must do now is carry that over to the other files. If you are picking out that A is "heavy" and B is "light" then you have something you are able to latch onto. Pick whichever you like, A or B and try and match that to the others. No rush. Just do one or two at a time and then come back to them.
I am flabbergasted, you are capable of identifying which hand the player used by listening alone, but you cannot say whether it is five or six notes!
This is absolutely stunning stuff, can you tell what the musician was thinking while he was playing?
Are you using expensive audiophile equipment?
Last edited:
I also listened in mono, and left channel only. Even with mono and left channel only, ABX results are consistent, with 5 files used alternatively in ABX. I only do not know WHAT I am discerning, when error information is higher in amplitude than usable sound difference. I am sure that the error is unintentional, that we are not intentionally cheated. But the error is there, and it dominates the usable differences between files.
I don't know whether my mono channel is left or right
(I don't know which part of stereo audio jack, the tip or the lower part, should go to left or right during assembly).It will be easier if we focus on the very high frequency only as the difference is there (I believe the single Shizuki cap is the one sample that has higher amplitude in high frequency).
IF for example error information ruins the characteristics consistency (i.e. files going through 5 caps should have the treble delayed or rounded) this can be proven later. How? When I'm sure that C is A based on similarity in the characteristics and Mooly reveals C is B then I will prove that C is not B by ABX 10/10 or even more. This will prove that the test is not valid, depends on the number of such occurrences.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.