[LAU] ABX test (was Re: What is the best MP3 encoder?)

[Peder Hedlund]

Quoting Ralf Mardorf <ralf.mardorf at alice-dsl.net>:

> audio ABX tests are absolutely correct, MP3s not only sound lossless,
> they improve the sound a lot. So lets spread this news and burn
> everybody who doesn't believe in this :p.

Despite your slight sarcasm an ABX audio test *is* correct.
For that particular song, that particular encoder and that particular  
listener.

If you score, say, a 5% guessing probabililty in a 10 run test that  
means that it's very likely you really *can* hear the difference  
between the two files and hence the encoder isn't transparent.
And, to clarify if it hasnt been made clear before, an ABX when it  
comes to testing a particular codec (lame, FhG, Xing) or encoder type  
(mp3, vorbis, opus, aac) really isn't "which file sounds better", it's  
"do I hear the difference between the two files".
If you *do* hear the difference it's obviously your prerogative to  
claim that the encoded file sound shite, even though the test in  
itself doesn't ask that question, but if your score is around 50% you  
really wouldn't have a leg to stand on then it comes to such a claim.

> I can't continue this discussion, I've to do some ABX tests to find out
> what washing agent makes white clothes the most white and what motor oil
> of the thinking motor oils has got the highest IQ.

Since you don't list any encoder test you really should try one once,  
just to be able to say "been there, done that, beat it".
If you look at the video (http://www.youtube.com/watch?v=jt7GyFW4hOI)  
it really isn't that complicated to set up and with your golden ears  
you should be able to fly through a 10 run test with a 0.1% guess  
probability in less than 2 minutes.

- Peder