Re: digital recording mystery

["Jonathan El-Bizri" <ssrndpty@hotmail.com>]

> I haven't thought it completely through, but I don't think lossy
compression
> works that way.  If it did, for instance, then each generation of 
>minidisc
> dubbing would significantly short the song.  In Petr's example of 0.4
> seconds shorter in a four minute song (with one minidisc generation), 
>then
> ten generations of dubbing would make the song 4 seconds shorter.  And I
> haven't heard anything about multiple minidisc generations shortening
songs
> in this fashion; only about the reduction in sound quality.
>

No, I know it works that way; I deal with lossy audio compression formats
all day long.

There are always discrepancies between files, even encording using the same
codec, at different settings. You won't always get less each time, it
depends on the source material - it's an artifact of the codec.

> Generally, lossy compression reduces the complexity of the signal and not
> it's timing.
>

I would assume the changes are caused due to anti-aliasing in the encoded
file - it would be unreasonable not to expect aliasing to occur, 
considering
what is going on.