|
Home
/ FAQ
/ News Classifieds / Events |
Audio Asylum Thread Printer |
Get a view of an entire thread on one page |
75.50.163.222
| '); } else { document.writeln(''); } } else { document.writeln(''); } } else { document.writeln(''); } } // End --> |
What do I hope to learn from doing comparisons of interconnects? I see three "levels of knowledge."
It will take many weeks to do this test, as I can only do two trials per day and likely only one day per week. I am depending on the generosity of a friend to help me connect/disconnect cables and don't know how many trials we will reach, or what my success rate will be in the end.
Right now my score is 2/2. Yes, that's right, I know my result after each trial, which increases the danger of cherry-picking the result by stopping after a good run.
- The first thing I wanted to know was: what happens, anyway, when I listen to a cable without knowing its identity? I did that a few times years ago, but my listening skills have advanced. I was curious to know: would my sighted impressions of cables disappear in a puff of smoke? Would all become confusion? Or would I have relatively the same impressions? (Answer below.)
- Second thing would be to convince myself the difference is real, based on the test results, but not adhering to a strict interpretation of them. For an extreme example, if I missed the first 6, but had a good run of 10/10 at the end, and I could feel myself getting better at the task, I might interpret the results as: "I needed practice, that's why I missed the first six, but I'll throw those out, and accept the last ten." Subjective experience would play a role here. If I felt like I was just plain guessing, I would be hesitant to conclude the difference is real, but if my experience really "felt" like "I knew what I was doing" I would probably accept fewer trials or cherry pick runs where I felt confident.
- The previous level of knowledge is still partly intuition and self-knowledge. So the third level is a rigorous study that can be peer-reviewed and accepted by the scientific community.
My goal is to get as far toward level 3 as I can before the patience of myself and my friend runs out.
So what happened when I listened blind? Note that in this test, A and B were randomly assigned and unknown to me, so I always had to go on sound alone, even when listening to the so-called "known cables." And what happened was that I had similar impressions as I do sighted. I noted different microdynamics, different apparent balance between registers, different levels of transparency, or midrange beauty, etc.
The only glitch was that I somewhat stupidly forgot to listen sighted (in a critical fashion) to either cable ahead of time! I compared an expensive Cardas cable to a Radio Shack cheapie. For some music, I preferred the Radio Shack! Afterward, I listened more closely and discovered that the Cardas cable seems to have an extremely rolled-off high-end compared to another cable of the exact same make and model. So I will use the other Cardas cable next week. I realize this is another statistical wrinkle ---for level 3 I can only count trials with the same configuration.
I'll start at the end. You are comparing an expensive Cardas cable to a Radio Shack cheapie. I have nothing against Radio Shack cheapies. I use them myself on my 2 subs. I like em actually.
But would any self-respecting golden-eared audiophile ever think that the Radio Shack was better than the Cardas? No, never. Me notwithstanding. So whatever prior statistical distribution you may have used for the Cardas and Radio Shack should have included golden-eared audiophiles, which you seem to be a part of. Which means you should have used a ONE-TAILED statistical distribution in evaluating it. In other words, the idea that Radio Shack was better than Cardas should have been ruled out beforehand. So any numbers without using this directional hypothesis are wrong.
Next, the permutation in which A and B are randomly assigned should be generated in an unbiased manner randomly. We just have your word here that this was true.
The total number of comparisons should be decided beforehand by a power analysis. Otherwise it is as you said, cherry-picking.
Are you evaluating "A is better than B", or "A is different than B"? See above, for the difference between one-tailed and two-tailed hypotheses.
"""But would any self-respecting golden-eared audiophile ever think that the Radio Shack was better than the Cardas? No, never. Me notwithstanding. So whatever prior statistical distribution you may have used for the Cardas and Radio Shack should have included golden-eared audiophiles, which you seem to be a part of. Which means you should have used a ONE-TAILED statistical distribution in evaluating it. In other words, the idea that Radio Shack was better than Cardas should have been ruled out beforehand. So any numbers without using this directional hypothesis are wrong."""
I don't know why you are referring to this in the past tense. The test hasn't happened yet (except for two initial trials which are really part of a learning and experimental period).
I don't pre-judge that the Radio Shack is worse than the Cardas. It may be different for different test tracks. I deliberately use different music each time I try to compare trials, as I did in my initial two trials, so as to keep my ears fresh and avoid a situation where I develop expectations too easily. In the test protocol, you'll notice that I don't know the identity of A and B. I don't know how often this is done in ABX tests, but I DO NOT want to know which of A and B is the expensive cable, to be sure I'm judging on sound alone.
As for one-tailed vs two-tailed distributions, you will have to explain more. I don't follow what you wrote.
"""Next, the permutation in which A and B are randomly assigned should be generated in an unbiased manner randomly. We just have your word here that this was true."""
My friend tossed a coin.
"""Are you evaluating "A is better than B", or "A is different than B"? See above, for the difference between one-tailed and two-tailed hypotheses."""
A is different than B. That's very important. To stay open to the possibility that preferences may change depending on the music I'm using, or even the day the test is run, I only try judge that A is different than B.
Have your friend say he changed the connectors and he really didn't.
And no peeping.
* I'm like a one-eyed cat peeping in a seafood store *
It's good to have an illusion, I guess - but it's just that, an illusion.
> And what happened was that I had similar impressions as I do sighted. <
Yes... just not each time in my case. But enough times to convince everyone involved.
> The only glitch was that I somewhat stupidly forgot to listen sighted (in a critical fashion) to either cable ahead of time! <
Then I'm not sure what your point was in testing. You just assumed you'd hear a difference blind even though you didn't try them sighted?
> If I felt like I was just plain guessing, I would be hesitant to conclude the difference is real, <
I guessed on a few trials but I notate when I'm not sure so I can backtrack and see what my mood was like, etc (I do my trials long term - 1 trial over the course of an evening) to see what happened. My guesses are almost always wrong.
> So the third level is a rigorous study that can be peer-reviewed and accepted by the scientific community. <
(snicker) - Good luck with that one! :)
At any rate, I'll be interested to see your results. There have been many positive blind tests on cables but also a lot of negative ones. I do not expect to be able to tell all cables apart. Indeed, many of them show no sonic differences to me even in sighted auditions.
> > The only glitch was that I somewhat stupidly forgot to listen sighted (in a critical fashion) to either cable ahead of time! < <
> Then I'm not sure what your point was in testing. You just assumed you'd hear a difference blind even though you didn't try them sighted? <
I did try them sighted, but not critically. In other words, I had the Cardas cable hooked up at one point, and at one point I had tried the Radio Shack. I was hardly even paying attention. For some reason, this made me sure they were different. They were very different, actually, but the Cardas cable was terrible. I suspect it's counterfeit. I got it from Audiogon.
> They were very different, actually, but the Cardas cable was terrible. I suspect it's counterfeit. I got it from Audiogon. <
I think you said the Cardas rolled off the treble. Some of them are designed to do that very thing but I can't recall which model(s). Seems to me one of them was a fairly low cost model (for Cardas) but there may be others. At any rate, it may be the real deal.
This cable is the Cardas Neutral Reference. I borrowed a Neutral Reference from the local dealer---same make model and length---and they sound nothing alike. The dealer's sounds gorgeous and high-resolution; the one I bought from Audiogon is extremely rolled-off, vague, fuzzy.
Ok, you're right, then. I used to own Neutral Ref and it was a killer cable - not rolled off at all.
Post a Followup: