Quote:
An encoder is NOT a standard. MP3 and audiophile should not even be regarded in the same sentence. How can you even say AAC is not a much better compression scheme than MP3? I also dislike how people measure what the codec throws away and then somehow can say that it's not musical information just by looking at its graphical representation. It's the same kind of crap as people saying anything greater than 16/44.1 is just adding non-musical information...
The MP3 standrad just tells you the format of frames, headers and so on, the psychoacoustic algorithms for deciding what information to discard are a matter of implementation, from tests done LAME is better at making those decisions than earlier FHG encoders, certainly early FHG encodes could be very poor, though this may have improved, I have not used FHG since 2000.
I use MP3 daily in my office rig and at home and in mobile settings, but I consider myself a music lover and not an audiophile. I have about $5K worth of kit and do not consider MP3 a waekness. For the vast majority of cases and the vast majority of listeners a modern LAME encode at VBR 0 or 320 is
perceptually transparent, but since you refuse to put your golden ears to the test no amount of rational discourse will ever persuade you on this.
The psychoacoustic algorithms decide what to discard using concepts such as
masking which models what frequencies will not be missed as they would not normally be heard in the context of the given signal, by and large this modelling is pretty good. Also a lot of high frequency data is discarded as it is less audible and at much lower levels than data in the 0 - 5K range.
As for 16/44.1 vs high res, well
there is no controlled listening test data to date that indicates that in normal listening that the difference between high res and red book encoding schemes is audible and there is a peer reviewed published paper that sugests that it is not normally audible. You may have more data in a high res encode but the information recreated on playback may not be effectively different, i.e the waveform reconstructed my not be functionally different or perceptually different.