amey01;655744 Wrote: > I don't think we need to turn this thread into a discussion about > digital theory. > > There is plenty of information about why 16/44.1 was never going to be > sufficient for quality audio (eg. filtering, anti-aliasing, distortion, > loss of low-level resolution and finally raw bandwith limitation, etc). > > > If you are genuinely interested (it is hard to tell in a forum > environment), I suggest you do your research. > > If not, then deaf-Freddy can hear that marked differences > high-resolution audio (or analogue for that matter) gives over 16/44.1.
I am really interested and yet I do not know why 16/44.1 was never going to be sufficient for quality audio.... I'm inclined to agree with Phil. If you are a real expert with a sound grounding in information theory and/or electrical engineering then I am about to make a fool of myself, but here goes.... The problem for most audiophiles wanting to learn about this stuff is where to learn it. It is perfectly natural to assume that people writing in magazines etc will a) know what they are talking about; and b) write frankly; c) be cautious and objective about anecdotal evidence etc In fact there is an awful lot of rubbish written in supposedly respectable publications- meaning pieces which regurgitate dubious nonsense. The long and short of it is that there are numerous myths and half truths in circulation amongst audiophiles. I believe that 1) the idea that 16/44 is inherently flawed it one of these, although I accept that if you wanted to be on the safe side you might have preferred a little more margin for error. 2) the idea that higher res formats are really better sounding to real people is doubtful. I do not pretend to have deep technical knowledge but I have taken the trouble to read some of the basic textbooks on information theory & electrical engineering, the orginal Shannon paper "Communication in the Presence of Noise", various pieces by Jim Lesurf, Dan Lavry, Daniel Weiss etc which I reckon is about as far as an amateur can be expected to go. I am not pretending to be an expert, but I do think I can generally understand what experts are saying as experts. I can't understand what is supposed to be the technical limitation of 16/44 which means that it cannot accurately reproduce what you can hear. AFAIK the following propositions are not genuinely contentious a) a continuous band limited signal can be perfectly reproduced from its samples (which is a mathematical truth) b) quantised samples having only a limited number of possible samples will with dithering allow perfect reproduction of that signal subject to the presence of noise with exactly the same character although much lesser magnitude than analog noise (ditto). c) the effect of dither is that signals below lsb in magnituide can be detected because they are greater in magnitude than the noise level in the adjoining region (not the total noise level across the whole spectrum which is irrelevant) d)that noise can be shaped and "thinned out" using oversampling to reduce its incidence in the areas to which the ear is most sensitive so as to increase the detectability of signals in the critical band even further (oversampling and noise shaping do this in different ways but the effect is similar) e) although analog filters would have a job doing this without phase and all sorts of other problems, most of the work can now be done digitally by purely mathemetical proceesses which will not vary with age, temperature conditions or anything else and can operate with vastly more than 16 bits precision even if the recorded samples are only at 16 bits. f) there is no evidence that the human ear can detect frequencies over 20ish kHz. g) if frequeines above 20kHz were to affected our hearing they would have to do so by modulating the frequencies below 20kHz in a way which would be captured by simply recording the frequecies below 20kHz [this bit might be contentious, but it is worth remembering that f) has implications and noone has produced any evidence AFAIK that frequencies above 20+kHz can be detected at all] i) all of the advantages of 16 bit plus and higher sampling rates (filter slopes, noise attenuation in detectable spectrum etc can be reproduced by oversampling etc by DSP in the DAC. They don't require the data to be at 24/96 or whatever and that is before we get onto the question of whether the transducers in the chain can reproduce anything above 20ish KhZ accurately (now say that again substituting 48kHz) Now what bothers me is that the supposed reasons why 16/44.1 was never gong to be enough ie "filtering, anti-aliasing, distortion, loss of low-level resolution and finally raw bandwith limitation, etc" sounds suspiciously like a list of audiophile myths based on misunderstanding of the above points. For example "loss of low level resolution" and "distortion" are notorious schoolboy errors, albeit ubiquitous ones. see http://www.audiomisc.co.uk/inadither/Page1.html I don't blame you for believing these myths, I did myself until I spent some time looking up how digital audio works. The problem is that it is not intuitive. IMHO If you start from the premise that there is something fundamentally wrong with 16/44 then listening to hi res enables you to get over this. It is probably like acupuncture: a very effective imaginary cure for imaginary illnesses. I have bought a good few hi res tracks and at the moment I am not at all sure I can tell them from dithered 16/44 versions of the same master. I am not certain because frankly, I recognise that if i listen to the same recording on the same equipment twice, it often strikes me differently. I can only say something is better if its sound goes beyond that surprising wide baseline variation (the thermal noise of subjective listening). I am aware that many experts have said that it might have been better if we had had 20/60 from the outset. But from what i can tell that was partly to do with margins of error and partly to with problems in analog filtering which no longer apply as a result of the availability of cheap powerful DSP chips making digtal filtering, oversampling and noise shaping easy-peasey. In reality 16/44 should be way more than enough for audio. Even if it were correct, it would not support the suggestion that anything more than 24/96 was required. The views of Dan Lavry and Daniel weiss can easily be found on the net. In short 1 16/44 is almost certainly "good enough" (a concept which is unfortunately anathema to audiophiles) 2 if it isn't then 24/96 is way over the requirement assuming a generous margin for error. I cannot also avoid adding that anyone who starts from the premise that analog recording is inherently superior to digital is IMHO barking up the wrong tree. -- adamdea ------------------------------------------------------------------------ adamdea's Profile: http://forums.slimdevices.com/member.php?userid=37603 View this thread: http://forums.slimdevices.com/showthread.php?t=89733 _______________________________________________ audiophiles mailing list audiophiles@lists.slimdevices.com http://lists.slimdevices.com/mailman/listinfo/audiophiles
