The 128/256k comparison is fairly worthless without being blinded so I thought I'd set up this blinded poll. Listen to the same file here and then vote on which ones sounded better.
One may only select a single choice when voting. This is what your categories should be:
(1) There is no noticable difference between either pairs
(2) There is no noticable difference between Mozart A/B, but REM A sounds better
(3) There is no noticable difference between Mozart A/B, but REM B sounds better
(4) Mozart A sounds better, but there is no noticable difference between REM A/B
(5) Mozart A sounds better and REM A sounds better
(6) Mozart A sounds better and REM B sounds better
(7) Mozart B sounds better, but there is no noticable difference between REM A/B
(8) Mozart B sounds better and REM A sounds better
(9) Mozart B sounds better and REM B sounds better
Unfortunately, it is possible to cheat the test by looking at the size of the m4a files (i.e. don't play them in the browser, but rather download them to your desktop and look at the file-sizes). Because of this, the outcome of the survey will be un-trustworthy (though each individual will know whether they had to cheat or guess to get the right answer). It was a good idea, though; and for the record, I couldn't tell the difference between the sound quality of either version.
And for people who are really good at self-delusion consider the following: if you suppose that there really is a difference between each of the files' bitrates then there are only 4 (not 9) possible outcomes. By guessing, you will be correct with probability 25%. It is, however, common for people to ignore their failures and focus on their successes. Because of this, it is quite possible to "feel" that you know what the better sounding files are, guess those choices, get only one correct (or possibly both), and then upon finding that one (or both) "guesses" were right, allow that to re-inforce your belief that you really did notice the difference. In this way, the probability is 75% that your choice will be re-inforced by one successful "guess." So, to all that would test themselves I would say, only if you "know" that one file is better than the other (i.e. you can point to specific audio artifacts in each track that indicates that one is superior to the other), then (and only then) do you really pass the test. Minus the cheating, all other "guesses" would likely be averaged out to no effect with enough people taking the test and your "feeling" that your correct guess meant that you really did notice the difference is probably just crap.
One may only select a single choice when voting. This is what your categories should be:
(1) There is no noticable difference between either pairs
(2) There is no noticable difference between Mozart A/B, but REM A sounds better
(3) There is no noticable difference between Mozart A/B, but REM B sounds better
(4) Mozart A sounds better, but there is no noticable difference between REM A/B
(5) Mozart A sounds better and REM A sounds better
(6) Mozart A sounds better and REM B sounds better
(7) Mozart B sounds better, but there is no noticable difference between REM A/B
(8) Mozart B sounds better and REM A sounds better
(9) Mozart B sounds better and REM B sounds better
Unfortunately, it is possible to cheat the test by looking at the size of the m4a files (i.e. don't play them in the browser, but rather download them to your desktop and look at the file-sizes). Because of this, the outcome of the survey will be un-trustworthy (though each individual will know whether they had to cheat or guess to get the right answer). It was a good idea, though; and for the record, I couldn't tell the difference between the sound quality of either version.
Yes, I realize it is possible to cheat but I would hope people here are more interested in a real result than skewing it one way or the other.
I have asked the mods, via the 'Edit' note to change the exclusivity of the voting to allow multiple choices but they have not done so yet (I hit the post button one instant too soon. Is there another way to contact the Moderator to get this fixed, or maybe its better to delete this thread and start another with the right settings?
Yes, I realize it is possible to cheat but I would hope people here are more interested in a real result than skewing it one way or the other.
I have asked the mods, via the 'Edit' note to change the exclusivity of the voting to allow multiple choices but they have not done so yet (I hit the post button one instant too soon. Is there another way to contact the Moderator to get this fixed, or maybe its better to delete this thread and start another with the right settings?
I would just start with a new thread.
And I don't think that you should be so confident in people's fairness. Some people are convinced (right or wrong) that 128AAC is crap. And there is a subset of these people that would classify as "true believers," by which I mean that they are so convinced that 128AAC is lacking that they will cheat so that others will accept their belief as "right." This is why researchers have to work so hard to design good double-blinded studies (which is what you're attempting to do, though only partially successfully -- not that this can be avoided in this case). The truth is that there are people who want the world to re-inforce what they already "know" to be true.
And I don't think that you should be so confident in people's fairness. Some people are convinced (right or wrong) that 128AAC is crap. And there is a subset of these people that would classify as "true believers," by which I mean that they are so convinced that 128AAC is lacking that they will cheat so that others will accept their belief as "right." This is why researchers have to work so hard to design good double-blinded studies (which is what you're attempting to do, though only partially successfully -- not that this can be avoided in this case). The truth is that there are people who want the world to re-inforce what they already "know" to be true.
Yes, I know. I work with experimental psychologists all the time. I guess I could re-code the 128 at 256 to get the same file size. I'll look at that before proceeding.
Yes, I know. I work with experimental psychologists all the time. I guess I could re-code the 128 at 256 to get the same file size. I'll look at that before proceeding.
That might not be the best idea. Re-encoding lossy to lossy would decrease (theoretically) the quality. If you want to try equalising the file sizes, try re-encoding both to AIFF (i.e. Apple lossless). The file sizes might, then, be uncorrelated with quality. Of course, the sizes will be large in this case, but you can't win them all.
Comments
(1) There is no noticable difference between either pairs
(2) There is no noticable difference between Mozart A/B, but REM A sounds better
(3) There is no noticable difference between Mozart A/B, but REM B sounds better
(4) Mozart A sounds better, but there is no noticable difference between REM A/B
(5) Mozart A sounds better and REM A sounds better
(6) Mozart A sounds better and REM B sounds better
(7) Mozart B sounds better, but there is no noticable difference between REM A/B
(8) Mozart B sounds better and REM A sounds better
(9) Mozart B sounds better and REM B sounds better
Unfortunately, it is possible to cheat the test by looking at the size of the m4a files (i.e. don't play them in the browser, but rather download them to your desktop and look at the file-sizes). Because of this, the outcome of the survey will be un-trustworthy (though each individual will know whether they had to cheat or guess to get the right answer). It was a good idea, though; and for the record, I couldn't tell the difference between the sound quality of either version.
And for people who are really good at self-delusion consider the following: if you suppose that there really is a difference between each of the files' bitrates then there are only 4 (not 9) possible outcomes. By guessing, you will be correct with probability 25%. It is, however, common for people to ignore their failures and focus on their successes. Because of this, it is quite possible to "feel" that you know what the better sounding files are, guess those choices, get only one correct (or possibly both), and then upon finding that one (or both) "guesses" were right, allow that to re-inforce your belief that you really did notice the difference. In this way, the probability is 75% that your choice will be re-inforced by one successful "guess." So, to all that would test themselves I would say, only if you "know" that one file is better than the other (i.e. you can point to specific audio artifacts in each track that indicates that one is superior to the other), then (and only then) do you really pass the test. Minus the cheating, all other "guesses" would likely be averaged out to no effect with enough people taking the test and your "feeling" that your correct guess meant that you really did notice the difference is probably just crap.
One may only select a single choice when voting. This is what your categories should be:
(1) There is no noticable difference between either pairs
(2) There is no noticable difference between Mozart A/B, but REM A sounds better
(3) There is no noticable difference between Mozart A/B, but REM B sounds better
(4) Mozart A sounds better, but there is no noticable difference between REM A/B
(5) Mozart A sounds better and REM A sounds better
(6) Mozart A sounds better and REM B sounds better
(7) Mozart B sounds better, but there is no noticable difference between REM A/B
(8) Mozart B sounds better and REM A sounds better
(9) Mozart B sounds better and REM B sounds better
Unfortunately, it is possible to cheat the test by looking at the size of the m4a files (i.e. don't play them in the browser, but rather download them to your desktop and look at the file-sizes). Because of this, the outcome of the survey will be un-trustworthy (though each individual will know whether they had to cheat or guess to get the right answer). It was a good idea, though; and for the record, I couldn't tell the difference between the sound quality of either version.
Yes, I realize it is possible to cheat but I would hope people here are more interested in a real result than skewing it one way or the other.
I have asked the mods, via the 'Edit' note to change the exclusivity of the voting to allow multiple choices but they have not done so yet (I hit the post button one instant too soon. Is there another way to contact the Moderator to get this fixed, or maybe its better to delete this thread and start another with the right settings?
Yes, I realize it is possible to cheat but I would hope people here are more interested in a real result than skewing it one way or the other.
I have asked the mods, via the 'Edit' note to change the exclusivity of the voting to allow multiple choices but they have not done so yet (I hit the post button one instant too soon. Is there another way to contact the Moderator to get this fixed, or maybe its better to delete this thread and start another with the right settings?
I would just start with a new thread.
And I don't think that you should be so confident in people's fairness. Some people are convinced (right or wrong) that 128AAC is crap. And there is a subset of these people that would classify as "true believers," by which I mean that they are so convinced that 128AAC is lacking that they will cheat so that others will accept their belief as "right." This is why researchers have to work so hard to design good double-blinded studies (which is what you're attempting to do, though only partially successfully -- not that this can be avoided in this case). The truth is that there are people who want the world to re-inforce what they already "know" to be true.
I would just start with a new thread.
And I don't think that you should be so confident in people's fairness. Some people are convinced (right or wrong) that 128AAC is crap. And there is a subset of these people that would classify as "true believers," by which I mean that they are so convinced that 128AAC is lacking that they will cheat so that others will accept their belief as "right." This is why researchers have to work so hard to design good double-blinded studies (which is what you're attempting to do, though only partially successfully -- not that this can be avoided in this case). The truth is that there are people who want the world to re-inforce what they already "know" to be true.
Yes, I know. I work with experimental psychologists all the time. I guess I could re-code the 128 at 256 to get the same file size. I'll look at that before proceeding.
Yes, I know. I work with experimental psychologists all the time. I guess I could re-code the 128 at 256 to get the same file size. I'll look at that before proceeding.
That might not be the best idea. Re-encoding lossy to lossy would decrease (theoretically) the quality. If you want to try equalising the file sizes, try re-encoding both to AIFF (i.e. Apple lossless). The file sizes might, then, be uncorrelated with quality. Of course, the sizes will be large in this case, but you can't win them all.