waltzingbear wrote:

Art Dudley is either trolling or hasn't been paying attention to the ADDA manufacturers as regards Nyquist and people not saying enough is enough. Dan Lavry (Lavry Engineering) has been arguing since the first 4X clocking converters came out that they were more than overkill and even detrimental to making the sound better.

Dan Lavry is swimming against the tide here. And having read what he has to say on the subject, I have to say that in my experience with high–sample-rate recordings, which I have been making since 1997, and given converters of equal quality, he is just plain wrong.

See my report on the blind test organized by Philip Hobbs of Linn at the 2007 AES Conference: http://www.stereophile.com/content/watching-detectives .

John Atkinson
Editor, Stereophile

Art's article portion on Nyquist gives me the wrong impression of what is really going on.

Art wrote that "The Nyquist Theorem suggests that,...". I thought that that theorem 'demonstrates' that a sampling rate at least as high or higher than the highest frequency of interest (Art's example of 20 KHz) will be mathematically reconstructed exactly. So a 44.1 KHz sample rate will reconstruct a signal, that has no frequencies above 20KHz, exactly.

Art wrote two paragraphs later that within 15 years of the CD introduction that others could "... hear and appreciate the improvements in sound reproduction associated with rising sampling rates - rates more than double those 'proven' by the Nyquist Theorem as sufficient to the task." Did the Nyquist theorem 'prove' that 20KHz was sufficient to reproduce recorded music? I think not, only that a signal containing frequencies up to half the sampling rate would be reconstructed perfectly. Someone linked the 20 KHz audio band to Nyquist.

The above statement made me wonder about Nyquist until I re-read what Art wrote. Then I realized that, unfortunately, as written, the logic jumps directly from the Theorem to 20 Hz-20KHz sound reproduction, with the implication that everything inbetween (recording equipment, software, reconstruction algorithms, playback equipment, etc.) is perfect, which it is not.
From the foregoing, a high sampling rate of 192 KHz will reproduce a signal containing frequencies up to 96 KHz. The Nyquist Theorem is not at question here.

The question is NOT what he concludes this section with, "Where's their messiah now?" I think the question that should have been asked is, "why does 96 KHz upper limit matter to sound reproduction as opposed to only 20 KHz?"