This has already been answered in bits and pieces, but I'll add my own just for kicks.
ralphpnj;331666 Wrote: > First, I think we can all agree that mp3 files, even 320kps mp3 files, > are different from flac files and different from the original wav file. So far so good. ralphpnj;331666 Wrote: > Now I'm told that when a properly conducted ABX is used to compare the > sound of a high quality 320kps mp3 file with the sound of a flac file > that it will be nearly impossible to reliably tell the difference > between the two files. That's the first mistake. Some files, when encoded at 320kbps, are quite easy to tell with a high degree of certainty from the lossless version. Yes, even in an ABX test. These tracks are called "killer tracks" and while some of them are artificially created specifically to trouble encoders, some of them are also real music tracks. I think we already have one person in this forum who has been able to make the distinction on a track in an ABX test, and I don't think this presents any problem. To the contrary, the rarity only helps prove my point. ralphpnj;331666 Wrote: > there are differences between the files but these differences can not be > heard, at least during the course of a properly conducted ABX test. That's what an ABX test would show, yes. ralphpnj;331666 Wrote: > Since the files are different and the ABX does not reliably reveal these > differences, the ABX test is flawed or, at best, incapable of resolving > the differences. That's also a possibility. For example, when the ABX test is done on a poor system, the listener is too drunk to pay attention, etc, etc. This doesn't mean that the whole scientific method is inherently flawed however. It's self-correcting. All anyone need do is point out the flaw in the test and the test can be re-run without those flaws, until everyone is satisfied. ralphpnj;331666 Wrote: > You and many other people agree with the first conclusion but, on the > other hand, I do not. I think that the second conclusion is the correct > one since the two files are different but the ABX test cannot reveal > these differences, so the ABX test is flawed. Not at all. In the second scenario, the burden is merely on the ABX tester to reveal their methodology and raw data (and that's easy enough to find on the thousands of tests on the Internet). Then the burden falls on the critic to identify the problem with the test, which can then optionally be re-run without the flaw. Science is nothing if not demanding. ralphpnj;331666 Wrote: > why...don't professional recording engineers record directly to 320kps > mp3 files First off, professional recording engineers need to use audio of an even higher quality that red book audio, so that when they mix and modify that audio (which are all lossy processes), the net result is still good quality. Secondly, as I've already pointed out, every now and then an mp3 really can be distinguished from lossless. This avoids the question entirely of how professional recording engineers after all this still manage to clip the hell out of their audio. ralphpnj;331666 Wrote: > So in the case of cables we are told if two items have the same > properties and test the same then they are equal but in the second case > we are told that although two objects are different and that even though > these differnces can even be verified by testing these two objects are > still equal. That's also easy to explain. With MP3s, just as with the high-end cables, it depends on what you're DOING with it. The electrical properties of various cables are just one of their properties, it just happens that this is the only property that's relevant to listening to audio. However, if I wanted to use a cable to suspend myself from the edge of a cliff, the high end cables may actually be better because they have more tensile strength. Analysing audio with a spectrum analyzer, or even just doing MD5 comparisons of the PCM data, is the equivalent of using an audio cable to go spelunking. As in, it has nothing to do with listening to audio. The only thing that you can do to show an audible difference is to listen to the audio. Use multiple listeners. Use high-end gear. I don't care, but you're never going to convince me that you don't have to listen to a track to see if it sounds the same. What we have here is perhaps a fundamental difference in credulousness. Let's say person A says they can tell the difference between a FLAC and an MP3 in an A/B test. Person B tells me that haven't actually heard either track, but their cat told them that they can't hear the difference. Person C tells me they can tell the difference between a FLAC and an MP3 in an ABX test. To me, this means that persons A and B are most likely imagining things. Person C has made an attempt to demonstrate that it's true, and if I accept his statement at face value, fine. If I doubt his results, I ask for his data and methodology and look for a problem. I think the audiophile world thinks person B is imagining things, person A proved something, and person C is some kind of masochist. -- CatBus ------------------------------------------------------------------------ CatBus's Profile: http://forums.slimdevices.com/member.php?userid=7461 View this thread: http://forums.slimdevices.com/showthread.php?t=51021 _______________________________________________ audiophiles mailing list audiophiles@lists.slimdevices.com http://lists.slimdevices.com/lists/listinfo/audiophiles
