max@uwm.UUCP (Max Hauser) (06/19/91)
Timothy L. Davis in article <13023@uwm.edu>: | | [Quoting Jon Brinkmann] | >I've been intrigued by advertisements for the _Chesky Jazz Sampler and | >Audiophile Test CD_. | > After nine Jazz selections, each a showcase for the company's | > proprietary 128-times oversampling technology, ... | ... | Unfortunately, the 128x-oversampled (Chesky-style) bongs are clipped ... | | My overall opinion: the jazz sampler section is as good as any other | sampler CD, and the audiophile test section could have been improved. | I'd like to see different tests ... less emphasis on their "wonderful" | 128x oversampling technology. There are many ways to do digital sampling | well, and this incredible oversampling is not necessarily the best way. ... I wonder if it will be useful to point out the following in order of mention. 1. 128:1 is one of the most common oversampling factors, even in the cheapest consumer audio products, and _per se_ it has very little to do with quality, positive or negative. 128:1 and 256:1 are the factors you end up using when you want to realize the full resolution of a 16-bit recording medium with the common noise-shaping-quantizer configurations. The invention of record of the youngest of these configurations was in 1969 and of the oldest, 1954. There may be something unusual about the way in which this particular firm performs its sampling process but if so it is not mentioned in the accounts quoted above; the factor 128 alone certainly does not imply something "proprietary" (rather the opposite). 2. 128:1 oversampling may seem "incredibly" high on first inspection but it is usually done for technical reasons having nothing directly to do with quality (often, in fact, it produces _cheaper_ sampling or reconstruction circuits, for a given electrical performance, than no oversampling at all -- you may not believe this if you don't know the field, but I can prove it if you care). The oversampling factor is sometimes then handed off to nontechnical marketing writers with blow-dry haircuts who proceed to milk the jargon for everything they can get. I don't mean to take anything at all away from the other thoughtful comments of these two posters (who may know all this) nor from what perceptual quality may be genuine in this firm's products. I just wanted to state these details for the record. Here as elsewhere facts like "128x oversampling" seem to have specious cachet and as someone involved in researching, popularizing and demystifying oversampling for a decade or so I try to do what I can. Max W. Hauser prls!max@mips.com {mips,philabs,pyramid}!prls!max Copyright (c) 1991 by Max W. Hauser. All rights reserved. -- "No matter how thin you slice it, it's still baloney." -- Rube Goldberg (UCB engineering alum), as quoted by Lee Roth in <1259@sousa.ltn.dec.com>