DCtoDaylight wrote: > The only caveat that I would throw into this, is to return the fact > that the Nyquist/Shannon limit is only mathematically correct for an > infinite number of samples. Without an infinite, or at least very long > series of samples, you cannot precisely define the amplitude of a sine > wave at a frequency just below the limit.
Being 22.1khz, and there is next to nothing over 15kHz in most real recordings. Granted, you do need a lot of samples, but 22k samples is a lot. Most FFT code uses blocks of about 100 samples (128, 256, etc) for movies, etc. > IMHO, 44.1kHz is on the edge for reproduction of 20kHz material, and > too low for 22kHz. There has been a fair amount of work that 19 or 20 bits @ 50kHz would have removed all the arguing. And avoided the evil brick wall filters used in the 80s for PCM stuff. What we fail to remember these days is that the RedBook spec was publicly released in 1980. At that time, a 1 megahertz computer was 20 feet long and cost a million bucks. Such a mainframe would have maybe 2 megabytes of ram. Moore's law has taken us a long way since then. Pat -- Pat Farrell http://www.pfarrell.com/ _______________________________________________ audiophiles mailing list [email protected] http://lists.slimdevices.com/lists/listinfo/audiophiles
