potential mp3 difficulties
Reply #23 – 2007-04-12 05:05:28
Pretty deep post there. I think the point of ABX testing is that: all things being equal (except the tester), the files should sound the same. The idealistic model human does not exist - at all!It seems to me, however, that it would be impossible to account for every variation in listening setup in psychoacoustic models. Exactly. Variations all around in abundance - hearing-abilities, speakers, room-acoustics, DSPs...... i fail to see how those factors differ in any way, regarding the topic at hand here. Any psychoacoustic compression has to have a reasonable safety-margin anyways. And it has to accept, that absolute perfection is impossible. There will be exceptions - the point is to make them so rare, that psychoacoustic lossy compression is "good enough" in the majority of cases. Thats why there is no such thing as "absolute transparency".... when i read posts on ha.org which say things along the lines of "transparency is binary", i can only smile and dismiss such naivity and lack of understanding about the background-theory behind psychology and even plain simply how critical realism works overally. Lossy compression is not perfect and there is no need for perfection - if you need something perfect, then use lossless - lossy compression is meant to be "good enough" in *almost* every case regarding normal listening. It has to be "good enough", not perfect. This is why neither so called "killer-samples" nor other extreme and unusual conditions show that an encoder failed - because such extreme and unusual conditions weren t the target anyways. - Lyx