[Bill Noble[_2_]]

"Steven Sullivan" wrote in message
...
Sonnova wrote:
On Fri, 21 Nov 2008 06:15:19 -0800, Steven Sullivan wrote
(in article ):

Kalman Rubinson wrote:
photograph.


If it takes up 3/4 of a spectrum, then that only means that the scale
of
the 'spectrum' has been absurdly expanded to ~100kHz.

Well, of course. Koschnike's point was that 192 KHz sampling has response
out
to 96KHz (half the sampling frequency). so, obviously, the DC-22KHz would
be
roughly 1/4 of a spectrum that goes to 100KHz.

Sorry, but what is the 'point' of making a point like that? All it is, is
an
entirely predictable confirmation of Shannon/Nyquist: your 'response'
will extend out
to just less than half of whatever your sample rate is. No one with a clue
would ever *expect* to see anything in spectral view beyond what the
Nyquist limit of 'response' dictates. So OF COURSE any spectral content
visible beyond 22 in 192 kHz-sampled audio, will be absent in a 44kHz
sampled version.


This shows that the LP faithfully preserves the HF content of the
master, while the CD does not.

snip-------

sounds like someone has a fundamental misunderstanding of sampled data
theory

if you are sampling a sine wave (or square wave) at a single frequency, so
long as your sample rate is 2X or greater than the fundamental, you will not
get ALIASING of the fundamental. This says nothing about distortion of
phase or waveform. If phase information is important, a significantly higher
sampling rate is needed - 10X is much more typical. A control system for a
large airplane, for example, had a roll off at 4 Hz - we found it necessary
to sample the input data at 60 hz to prevent phase induced instability.
That's 15X the maximum passband frequency.

I see no reason why this kind of effect does not apply at audio frequencies
as well.

[Eeyore]

Bill Noble wrote:

if you are sampling a sine wave (or square wave) at a single frequency, so
long as your sample rate is 2X or greater than the fundamental, you will not
get ALIASING of the fundamental. This says nothing about distortion of
phase or waveform. If phase information is important, a significantly higher
sampling rate is needed - 10X is much more typical. A control system for a
large airplane, for example, had a roll off at 4 Hz - we found it necessary
to sample the input data at 60 hz to prevent phase induced instability.
That's 15X the maximum passband frequency.

I see no reason why this kind of effect does not apply at audio frequencies
as well.

Because the ear is insensitive to phase info at the relevant frequencies.

Graham

[Steven Sullivan]

Bill Noble wrote:
"Steven Sullivan" wrote in message
...
Sonnova wrote:
On Fri, 21 Nov 2008 06:15:19 -0800, Steven Sullivan wrote
(in article ):

Kalman Rubinson wrote:
photograph.


If it takes up 3/4 of a spectrum, then that only means that the scale
of
the 'spectrum' has been absurdly expanded to ~100kHz.

Well, of course. Koschnike's point was that 192 KHz sampling has response
out
to 96KHz (half the sampling frequency). so, obviously, the DC-22KHz would
be
roughly 1/4 of a spectrum that goes to 100KHz.

Sorry, but what is the 'point' of making a point like that? All it is, is
an
entirely predictable confirmation of Shannon/Nyquist: your 'response'
will extend out
to just less than half of whatever your sample rate is. No one with a clue
would ever *expect* to see anything in spectral view beyond what the
Nyquist limit of 'response' dictates. So OF COURSE any spectral content
visible beyond 22 in 192 kHz-sampled audio, will be absent in a 44kHz
sampled version.


This shows that the LP faithfully preserves the HF content of the
master, while the CD does not.

snip-------

sounds like someone has a fundamental misunderstanding of sampled data
theory

if you are sampling a sine wave (or square wave) at a single frequency, so
long as your sample rate is 2X or greater than the fundamental, you will not
get ALIASING of the fundamental. This says nothing about distortion of
phase or waveform. If phase information is important, a significantly higher
sampling rate is needed - 10X is much more typical. A control system for a
large airplane, for example, had a roll off at 4 Hz - we found it necessary
to sample the input data at 60 hz to prevent phase induced instability.
That's 15X the maximum passband frequency.

I see no reason why this kind of effect does not apply at audio frequencies
as well.

Sounds like someone wants to challenge Shannon/Nyquist, again....

All you have to do , is show 'this kind of effect' in an audio sample.

--
-S
I know that most men, including those at ease with problems of the greatest complexity, can
seldom accept the simplest and most obvious truth if it be such as would oblige them to admit
the falsity of conclusions which they have proudly taught to others, and which they have
woven, thread by thread, into the fabrics of their life -- Leo Tolstoy

[Arny Krueger]

"Bill Noble" wrote in message


if you are sampling a sine wave (or square wave) at a
single frequency, so long as your sample rate is 2X or
greater than the fundamental, you will not get ALIASING
of the fundamental.

Sure you will. A square wave will contain significant harmonics of the
fundamental. For example, a 22.00 KHz square wave will pass your criteria in
a system sampled at the typical 44.1 KHz. The next harmonic will be the
third harmonic at 66.00 Hz. Since aliasing repeats itself in successive
mirror images as frequency increases, 66 KHz will be aliased to some
frequency above 20 KHz.

It is therefore an absolute requirement that any broadband input signal be
low-pass filtered so that it does not have significant content Nyquist.

This says nothing about distortion of phase or waveform. If phase
information is important,
a significantly higher sampling rate is needed - 10X is
much more typical.

Ironically, a 10x safety margin is not a requirement for a digital system.
While a digital system cannot do *anything* useful with a frequency at or
above the Nyquist frequency, there is nothing inherent that says that
frequencies just below the Nyquist frequency must have phase distortion.
It's all about the anti-aliasing filter, and if that filter is a oversampled
digital filter, all sorts of counter-intuitive things can be done.

A control system for a large
airplane, for example, had a roll off at 4 Hz - we found
it necessary to sample the input data at 60 Hz to prevent
phase induced instability. That's 15X the maximum
passband frequency.

I'm somewhat familiar with control systems - I did post-graduate work on
them. There are too many missing pieces in this anecdote to diagnose the
problem, but there is no need to sample at 15x in order to have minimal
phase error.

I see no reason why this kind of effect does not apply at
audio frequencies as well.

http://www.pcavtech.com/soundcards/L...644-xfus10.gif

shows the phase response of a high quality audio interface operating at a
44.1 kHz sample rate. The phase error was less than 5 degrees at 20 KHz.
A 3 KHz it was something like a degree and a half. I would not consider a
control system with a 2 degrees or even 5 degree stability margin to be
suitable for use in an airplane.

[Sonnova]

On Sun, 23 Nov 2008 20:23:45 -0800, Bill Noble wrote
(in article ):

"Steven Sullivan" wrote in message
...
Sonnova wrote:
On Fri, 21 Nov 2008 06:15:19 -0800, Steven Sullivan wrote
(in article ):

Kalman Rubinson wrote:
photograph.


If it takes up 3/4 of a spectrum, then that only means that the scale
of
the 'spectrum' has been absurdly expanded to ~100kHz.

Well, of course. Koschnike's point was that 192 KHz sampling has response
out
to 96KHz (half the sampling frequency). so, obviously, the DC-22KHz would
be
roughly 1/4 of a spectrum that goes to 100KHz.

Sorry, but what is the 'point' of making a point like that? All it is, is
an
entirely predictable confirmation of Shannon/Nyquist: your 'response'
will extend out
to just less than half of whatever your sample rate is. No one with a clue
would ever *expect* to see anything in spectral view beyond what the
Nyquist limit of 'response' dictates. So OF COURSE any spectral content
visible beyond 22 in 192 kHz-sampled audio, will be absent in a 44kHz
sampled version.


This shows that the LP faithfully preserves the HF content of the
master, while the CD does not.

snip-------

sounds like someone has a fundamental misunderstanding of sampled data
theory

if you are sampling a sine wave (or square wave) at a single frequency, so
long as your sample rate is 2X or greater than the fundamental, you will not
get ALIASING of the fundamental. This says nothing about distortion of
phase or waveform. If phase information is important, a significantly higher
sampling rate is needed - 10X is much more typical. A control system for a
large airplane, for example, had a roll off at 4 Hz - we found it necessary
to sample the input data at 60 hz to prevent phase induced instability.
That's 15X the maximum passband frequency.

I see no reason why this kind of effect does not apply at audio frequencies
as well.


Well, it MAY. But at this point it is an unknown because we are dealing with
a human sensory perception that's difficult, if not impossible, to measure or
even easily observe (unlike the the control system of an aircraft). Many
people contend that CD contains all of the information necessary for the
perfect human perception of music. Others say that CD hasn't the resolving
power and that higher bit-rates and more bits are necessary, or that a
different encoding scheme (such as DSD) are needed in order to capture the
important essence of music digitally.