[[email protected]]

On Aug 29, 1:56 pm, ScottW wrote:
On Aug 29, 10:37 am, John Byrns wrote:
Exactly what is the CD sample rate not high enough for, and how do you
justify that statement? If the audio is first run through a 20 kHz
brick-wall filter the CD sample rate is plenty high. Brick-wall filters
can be built with today's digital techniques that do not contribute
phase distortion to the filtered signal.

How does one implement an anti-aliasing filter on the
input before a signal is digitized.....digitally?

Yes, in essence.

Put in a VERY gentle analog filter whose low-pass is at,
say, 64x that of your base sample rate. Oversample the
A/D at 64x your final sample rate, e.g., 2.822 MHz, and
then do your real anti-aliasing filtering entirely in the
digital domain. You can now build yourself a very nice
filter at 20+ kHz that has in-band response anomolies
of under +-.1 dB, excellent phase repsonse and more.

Same thing for the output reconstruction filter: oversample
the stream, do your filtering digitally, then convert and follow
with a final very gentle analog filter. Those analog filters
now need not be at the top of the bandwidth, they can be
32x higher, far above the 5 times limit that Ludwig is
wagging about.

It, by the way, has been routinely this for the last couple
of decades in one form or another.