adrian@eagle.UUCP (A.Freed) (09/24/84)
This is a brief attempt at some sensible discussion on Vinyl and CD recordings, from someone involved professionally in related things for some time: Places where problems can be introduced: A/D conversion (filtering and sampling) unless you are computer synthesising your sound directly, (and few of us are) some conversion process has to be performed. Designing and measuring high quality A/D conversion systems for audio is not understood by many engineers yet. There is no perfect anti-aliasing filter! You can never do better than the original source. Alot of CD sources are analogue masters, and even worse copies of copies of masters. Interpolation on bad CD data as far as I can tell, unless you go to a lot of trouble to damage the surface of a CD, most CD machines won't ever have to invent sound for you, to make up for uncorrectable erroneous data. D/A conversion and filtering on output This problem is also new to most engineers. The most important way in which CD players differ as far as the sound is concerned is the characteristics of their anti-aliasing filters. This business of one or two DACS is a red herring. Fixed phase shifts are irrelevent. Frequency dependent shifts are the problem. There is also the problem of the non-linearity of the output of a DAC. The output can never be a step function (infinite energy would be required). Careful correction has to be done to make sure any distortion components are minimised or moved to a filterable frequency. There are all sorts of other conventional noise and distortion problems from the rest of the analogue chain to the output. From what I have read, few people are measuring any of these new performance parameters. Manufacturers are not publishing specifications of them and people really can hear the difference! One of the reasons are that they are not widely understood. Another is that they are not easy to measure.Even when they are published, there is not enough psychoacoustics research for us to make deep conclusions. Listening to your favourite music on machines that other people have recommended and that will be well serviced, is still the best way of choosing HIFI gear. People are confusing the near perfect medium, with the far from perfect reproduction process. (a problem that reproducing organisms have to!). The following things excite me about CD: The possibility of storing data about the music on the disk for manipulation by home computer. Being able to make perfect copies of what are not far from masters (when the recording industry is totally digital). There are lots of legal problems with this, but I am fed up with scratched disks and cassette tape. The Future It may be a depressing thought but there will probably be several CD formats. I would like a smaller one for a walkman (Sony have a prototype CD walkman). I would also like a larger capacity format for encyclopaedic collections. There will also be (some day) an integrated format for video/sound/data. Why don't we just post revues of CD's we have enjoyed and problems with CD machines? The new CD magazines are not giving good revues of enough disks, nor will they flame manufacturers who advertise in their rags!
newton2@ucbtopaz.CC.Berkeley.ARPA (10/05/84)
I very much liked Adrian Freed's posting on the practical engineering aspects ofCD equipment, as well as the reasoned segregation of the inherent appealing virtues of a discrete storage medium from the constantly shifting vagaries of the recording/reproduction gadgets currently available. I'd like to see more discussion of this sort, on this plane, about many of the topics that exercise so many of us nutty audiophiles. More care in isolating and identifying what are the virtues we are seeking to maximize when we tout or trash some gismo or technology. For example, can anyone get a handle (epistomologically) on what this Sheffield person seems to be saying (in a recent squib in Studio Sound) when he asserts that some of the digital demons can be exorcised by recording CDs that are dubs of (his) analog masters? Does anyone else claim that "transistor sound" can be laundered away by passing the signal through a triode? Seemingly yes, since many people assert that the "improvement" of some element in a concatenated chain of equipment can eliminate crippling inadequacies caused by the nature of the replaced element, despite the presence of doezens of identical elements up and down the chain (transformers, copper wire with merely tinned connectors, class AB amplifier stages, what-have-you). Oh well, "there I go again".