[RRod]

   
Well, equality is the point. If the waveforms are different, you start the subjective thing that will lead to nothing. 
Then take lossless 320 and 4000. If that is not equal, you have to admit that someone may hear differences.
 
 
I took bishots test. Please read the thread.

 
Equality is not the point, audibility is. And I saw you took his test. His conclusion of the results was that above 256k things got iffy for you. That's pretty much how everyone else is.

 

[mironathetin]

   
Equality is not the point, audibility is. 

No, you cannot prove audibility. You can only make statistical tests. And that is the raison d'etre of gigabytes of discussion in this database.

 

[cjl]

   
If you care to prove that 320 kbit mp3 is the same as 4000 kbit (lossless, mp3?), overplot both waveforms and show it. End of discussion. That the same waveform will drive speakers in an identical way should be easily acceptable. If you cannot plot, a correlation function that shows 100% correlation should be fine too.

Given that the differences are down at the -50-60dB level or so for most samples, chances are if you overplotted both waveforms, you wouldn't be able to visually discern the difference.You'd need to do a plot of the difference between the two, and magnify the scale immensely (or use a log scale or something like that) to actually see the differences between the waveforms.

 

[maverickronin]

  No, you cannot prove audibility. You can only make statistical tests. And that is the raison d'etre of gigabytes of discussion in this database.

 
All empirical knowledge is probabilistic.
 
If you want to take it your definition of "proof" to that level then you you will be unable to prove anything in the real world, including whether the waveforms are identical or not.  The Architect might just be screwing with you behind you back.

 

[RRod]

  No, you cannot prove audibility. You can only make statistical tests. And that is the raison d'etre of gigabytes of discussion in this database.

 
maverickronin already said what I was going to. But also remember that people like FB are stating that mp3 sounds *horrible*. Not "I can tell a difference in an ABX if I do really quick switching", but so horrible that it's obvious. He's supporting this notion by 300 ≠ 4000. If you want to accept that as reasoning, fine. I'll take actual ABX tests done with a low false positive rate over such reasoning any day.

 

[bigshot]

  No, you cannot prove audibility. You can only make statistical tests.

 
To determine audibility, you compare measurements to the established JDD (just detectable difference) threshold.
 
To establish the JDD threshold, you do double blind listening tests.

 

[cjl]

   
To determine audibility, you compare measurements to the established JDD (just detectable difference) threshold.
 
To establish the JDD threshold, you do double blind listening tests.

Alternatively, you could bypass the intermediate step and simply do a double blind listening test of the gear, file, or item in question. Either way, you can definitely prove audibility. You can't really prove inaudibility though, other than to simply say that tests have failed to show an audible difference. This is why the onus is on the person claiming an audible difference to demonstrate it.

 

[bigshot]

Ever notice that the people who claim to hear things that shouldn't be audible are the same ones that say they don't have time or interest in doing controlled testing?

 

[cjl]

  Ever notice that the people who claim to hear things that shouldn't be audible are the same ones that say they don't have time or interest in doing controlled testing?

It's remarkable how often that happens, even when they claim that it's really easily audible and obvious (in which case, a controlled test should take very little time and be really easy).

 

[earthpeople]

Tests are clearly just a way to trick their senses. They know what they hear!
 
 
 
/s

 

[Terja]

If you're starting a test with the idea that it 'shouldn't be audible' the test is already flawed. I see two kinds of biases going on here . . .

 

[bigshot]

I actually agree in a sense. There really isn't much sense in testing something you know won't get you anywhere. But people continue to reinvent the wheel and do tests to answer questions that were asked and answered long ago. I guess if they enjoy it, that's OK. But there are bigger fish to fry than "Do cables make an audible difference?" or "Can I hear the difference between redbook and high bitrate files?" Those ships have already sailed whether people want to admit it to themselves or not.

 

[earthpeople]

  If you're starting a test with the idea that it 'shouldn't be audible' the test is already flawed. I see two kinds of biases going on here . . .

 
But that's precisely why those who think they can hear it should be doing these tests. 
You're right that if you go into it already thinking there's no difference, the test isn't going to work properly. 

 

[bigshot]

  You're right that if you go into it already thinking there's no difference, the test isn't going to work properly. 

 
Only in the case of a VERY subtle difference. If the difference is clearly audible, even not believing there is a difference won't make any difference. If you can hear it, you can hear it.

 

[castleofargh]

 

  Besides, and opinion is something like "I prefer blue to green." A statement like "320 mp3 sounds like crap all the time" is NOT an opinion, because there are theoretical and practical arguments, and data, that can prove it untrue, or at least highly improbable. It gets even worse when the statement is justified by meaningless tautologies like "320 ≠ 4000." Why should such things get any air to breathe?

 
If you care to prove that 320 kbit mp3 is the same as 4000 kbit (lossless, mp3?), overplot both waveforms and show it. End of discussion. That the same waveform will drive speakers in an identical way should be easily acceptable. If you cannot plot, a correlation function that shows 100% correlation should be fine too.

here you're typically playing bookman's game. stop using those values, they're nothing, at best they are real values with different units. it started with 4k for screen resolution and somehow got mixed up with lossy mp3 being 320kb and why not lossless being 4000kb. but even then, if PCM is given as bit/sample rate and mp3 is given with kb, it's to show that they aren't the same tech, and only PCM should be compared to PCM in that way. kind of obvious why. it's plain wrong to even try making a comparison. but of course that doesn't stop him even though I already explained that to him some times back. he didn't understand, didn't look it up or simply just doesn't care if what he says is true.
 and nobody ever tried to say that mp3 was the same as lossless, only him pretends like we do. and that's a strong strawman argument, nothing more.
then he goes on "proving" his point with fallacies. that's what makes us react like we do. not that some guy doesn't share our opinion, but all the nonsense he doesn't mind using to make a point. you can only take him seriously for so long. I stopped long ago when he was talking about how dither was ruining the soundstage. the subjects change, but not his absolute certainty of being right while using incomplete knowledge.
 
and no matter how many times I and others already wrote that we never thought mp3 to be as good as lossless, his answers are always telling that we do.

shouldn't we be the ones offended here?
if all that crap cleared out, then after we could come to discuss what test is good enough or not, and the best way to discriminate mp3 from lossless. but it would be real nice to start by arguing about a subject that he didn't made up.
I went to length the other day to explain to him why I used mp3 on portable devices for space, for price, and for longer battery. nothing to do with sound quality. do you think he cared about that? no he went back to square 1 and blamed people for not hearing that mp3 is crap and that a 4K screen is better than a smaller one and stuff like that.