Fw: Letters To The Editor In Today's Globe

[rogerkirk]

"Dan Strassberg" wrote:
>HD Radio's lossy-compression algorithms are more advanced than MP3's 
>compression algorithms (which, IIRC,are not lossy). Although the point 
>is subject to heated debate and the conclusions about audio quality 
>are intimately related to the program content, some people say that 32 
>kbps in HD Radio produces audio quality that is comparable to that of 
>a 128-kbps MP3 stream.

.mp3 encoding is lossy by definition and by design.  What is "thrown away" is information that alledgedly will not be heard by the ear - for example a soft sound masked by a louder one (a simplification, I know.)  Unfortunately, that depends on whose ear is listening and on the complexity of the program material being encoded.  It has been postulated that possibly as many as one in ten people can detect .mp3 loss more readily than the other nine.

I did some empirical testing on .mp3 encoding at 128, 160 & 192-kbps.  At 128-kbps, there were frequently noticeable artifacts - especially
with cymbals and those petite glissando-chimes (don't know the real name) that are popular in MOR and smooth jazz.  At 192-kbps they become harder to discern and at 256 or 320-kbps disappear (for all practical purposes).  

Variable Bit Rate (VBR) allows the encoding bit rate to rise and fall to match the complexity of the encoded sound - while reducing the overall space required to store it.  However, I suspect that is not feasible with HD-radio which I would guess is Constant Bit Rate (CBR).