Quite a few people here appear to disagree with this statement. [ie sighted listening is useful and has some relationship to blind listening.]
Including Floyd Toole and Sean Olive. Let’s see if you prove them wrong, below.
That "sighted listening has some relationship to blind listening" is fully supported by the Harman listening data. Fig.17.12 in Toole's book, which he chose to use to illustrate this relationship,
That’s not what Toole says he was illustrating with that chart. He says the chart illustrates that “what listeners see changes what they (think) they hear.” So let’s use the chart to examine
that statement.
and which I discussed in detail in an earlier post in this thread, shows this:
blind: A=B > D=C
sighted: A=B > D>C
where = means that the ratings of the two speakers were within statistical errors (ratings difference of 0.3 or less).
All ratings were in the range 5.5 to 8, so just for fun I generated 4 random numbers in this range as the ratings of the 4 speakers. This was the result of doing this five times:
randomly generated ratings:
A=D > B=C
C > A=D > B
A > B=C > D
B > C > A > D
C > B=D > A
OK, and I will assume that you applied the same error bar size to assign a “=” etc.
So you can see that, if there were no relation between blind and sighted listening, it would be very unlikely for the sighted ratings to correspond as closely as they do to the blind ratings. (There are various statistical methodologies to quantify just how unlikely this is, but I'm not going to go through that exercise.)
An interesting presupposition of your analysis is that the sighted listening scores need to be completely random for there to be
no relation.
Look at the first two columns A and B (black, white) in Fig 17.12. Those speakers are visually identical, and their sound waves are statistically indistinguishable. Remarkably, when sighted, their
relative scores compared to one another are exactly the same as when blind. Why? Because listeners are effectively still blind in relation to the difference between those two speakers, so the blind test result is replicated wrt difference. I don’t think you gave consideration to the fact that A vs B is blind both times.
You are mistaken if you think that I and others think that sighted listening leads to random scores. (If that is your test, then it’s a straw man.) It’s just that, when sighted listening factors come into play, they are strong enough to dominate, ie “change what people (think) they hear”, and that is too big an issue to proceed on the basis that sighted listening is telling us what we need to know about the sound waves themselves.
So the only debate that is reasonable to have is whether this unquestionable relation between blind and sighted listening is "useful" or not.
Wait a minute. You misquoted
@tmtomh when you said people disagree with his statement and then quoted it as “sighted listening has some relationship to blind listening”. He actually said “sighted listening
is useful and has some relationship to blind listening”.
To dispute his statement, you have to dispute it as a logical AND statement, that both attributes are true simultaneously.
Secondly, even if it were assumed that we do have “a relationship” between sighted listening impressions and blind listening impressions, it won’t help us if it is not useful.
And that’s where it all comes crashing down, since you admit below that ‘useful’ is undefined. A broken clock has an “unquestionable relation” to the time of day, and I could claim that it is actually more useful than sighted listening impressions, because unlike sighted listening, we can predict when and how often it will be telling the truth. But still not useful enough to rely on it as a timepiece.
A key attribute of utility is
reliability. We can rely on a well-controlled listening test to be about the perceived attributes of the sound waves themselves, but we cannot rely on a sighted listening test to do that. Everyone has unique cognitive biases including unconscious biases, and like
@tmtomh pointed out, they are fluid in time (short and long term). How is one supposed to
rely on that? Even if you worked out and documented “a relationship” between your own personal sighted listening impressions and your blind listening impressions…it might not be the same tomorrow, or next month. Or it might. Useful?
And sighted biases are so multi-faceted, so many possibilities, that they can’t all be covered. Some are near-universal, like a preference for large speakers (and if you brought the same analysis to Toole’s chart for that factor you would conclude it is an “unquestionable relationship”),
and yet, you might have a personal bias in favour of small speakers because you are aware of their typically stiffer cabinets and lower baffle diffraction, and so your sighted listening leads to the opposite relationship to the one Toole’s chart indicates. Useful relationship? Reliable?
Here is a scenario that I have read on numerous occasions by audiophiles: you read a very positive review of a speaker and you go and audition it. You verify that you hear much of the attributes reported in the review (evidence that sighted listening is insightful about the sound waves?), and you love what you are hearing enough to buy the speaker. Your experience at home is splendid. Then you read another review of it by a reviewer who you hold in deep regard, but he describes a very irritating sonic aspect. Now you start hearing that sonic aspect at home too! And once you hear it, you can’t
unhear it. (
More evidence that sighted listening is insightful about the sound waves?)
Analysing the scenario, what aspect of the sighted listening impressions is
reliably correlated with the sonic attributes of the sound waves themselves? Probably
none. But were
any of them actually in the sound waves? Possibly…probably…maybe…some? Which ones? Dunno. Can we rely on those (presumed) correlations? Er, no. Useful? Definitely not.
As you can see, I am not saying that it is impossible for “a correlation” to exist, but it just isn’t useful, because it can be overridden at the drop of a hat.
I think it is. Matt thinks it is. Kal thinks it is (as confirmed by his taking the time and effort to do sighted subjective reviews). Many here think it is not.
Ask Kal if his work would get published in Stereophile if he omitted the sighted subjective reviewing component.
Since "useful" does not have precise definition, there is no way to settle this argument definitively.
I’m very disappointed to see that you are not completely persuaded by Matt’s insistences.
I’m going to restate from
a post I made in this thread only a few days ago, apologies.
- The critical factor is that the listening tests must be blind or double-blind - sighted tests cannot be trusted. - Floyd Toole
#285
- If you want to obtain an accurate and reliable measure of how the audio product truly sounds, the listening test must be done blind. - Sean Olive
https://seanolive.blogspot.com/2009/04/
Are you sure that your resistance to this authoritative conclusion on the matter is not, possibly at a level of unawareness, underpinned by a feeling that your sighted listening experiences seem just too real to
not be ‘about the sonic attributes of the sound waves themselves’?
cheers