Path: utzoo!news-server.csri.toronto.edu!cs.utexas.edu!sdd.hp.com!hp-pcd!hplsla!bryanh
From: bryanh@hplsla.HP.COM (Bryan Hoog)
Newsgroups: comp.dsp
Subject: Re: A question about the Nyquist theorm (CD sampling/filtering)
Message-ID: <9360020@hplsla.HP.COM>
Date: 8 Mar 91 04:04:08 GMT
References: <17510@milton.u.washington.edu>
Organization: HP Lake Stevens, WA
Lines: 23

>
>Some of the first ones used 1 D/A and sample and holds. The 
>BBC wanted to broadcast monophonic off of some CD's.  With
>the half sample time delay, it made the signal sound terrible.
>

   Let's see.  A half sample delay is 11.34 uSec.  Assuming they
 just added up the left and right channel to get mono, they
 essentially created a simple time delay filter.  This lowpass 
 filter would have a zero at 44.1 KHz.  The 3 dB frequency would be  
 22.1 kHz.  At 10 kHz, there would be about .6 dB of attenuation.
 This filter is linear phase.

   It sounded terrible?  I wonder what the mechanism was, since the
 L+R signal shouldn't be affected much.  But wait, the L-R signal
 is no longer cancelled completely.  It probably has a high pass shape
 that's the inverse of the L+R lowpass shape.

   But wait another second.  If the microphone that recorded the 
 material in the first place had been shifted a fraction of 
 an inch . . .

   Bryan Hoog