Digital BW, The Print

Any analogue filter (since they are "noncausal") will (mathematically) 
have the effect of a nonlinear phase response - which can be trainslated 
to different delays for different frequencies since the derivative of 
phase (group delay)  is related todelay of a frequency component through 
the filter. This description makes sence if you are only concerned with 
an input of two discrete frequencies. However, for a complex input - 
such as music - this description breaks down somewhat and usually goes 
under the name of phase distortion. BTW this same phase distortion is 
why the need for equalization using training sequences on phone line 
modems to get any decent bit rate over a phone line. By using wider 
bandwidth filters and higher samping rates, then the effect of nonliner 
phase distortion can be minimized by putting the passband of interest in 
the flat phase repsonse of the filter - hence all frequencies are 
"delayed" by the same amount.

Then by using digital signal processing a filter of constant delay over 
the passband of interest can be used. It is very easy to get filters of 
constant delay using digital filters and (mathematically) impossilbe to 
get this using analogue filters.

I would expect on the initial CD's there were two issues, the first 
would be the amplitude response - different frequencies have different 
weights on the output amplitude. Part of quality of music (or voice for 
that matter) comes from the harmonics in the signal. If these are 
weighted differently then it upsets this harmonic relationship. Then add 
to the phase effect you noted, this causes an inteference relationships 
that will have the effect to produce an amplitude modulation or "beat" 
between harmonically related frequencies- and this can be quite annoying.

The newer processing in the production should have solved these issues.

Truman


Austin Franklin wrote:

>I don't believe any conclusion can be drawn by that.  When CDs first came
>out, they did not use oversampling to smooth out the output signal, and that
>is why they sounded "digital", and in fact, sounded annoying to people who
>weren't mental patients as well.  This issue has long been attended to by
>the use of oversampling and first order filters.  Prior to using
>oversampling, third order filters were primarily used, and these induced
>phase shift in the output, simply because different frequencies take
>different times to travel through the filter.  The result is typically
>annoying to human hearing.  Not so with first order filters.
>
>Austin
>
>  
>