Audio compression reversible? - was Re: Apple Computer changes its name

["Alex Freed" <alexf@mirrow.com>]

"Matthew T. Russotto" <russotto@grace.speakeasy.net> wrote in message 
OMWdnfxCHaakwTDYnZ2dnUVZ_sudnZ2d@speakeasy.net">news:OMWdnfxCHaakwTDYnZ2dnUVZ_sudnZ2d@speakeasy.net...
> If you use exactly the same codec with exactly the same settings, I'd
> hope there would be no quality loss.  I don't know if this is the
> right term, but I call this a "stable" codec -- one where C(D(x)) = x

I doubt it.

While I don't know what kind of algorithms are used, most LOSSY ways
of compressing image or sound data rely on applying transcendental 
functions.
For example a descrete cosine transform. Time domain samples are transformed
into frequency domain. Then thing that are considered unimportant are left 
out.
However even if this step is not taken, there is always a rounding error as 
the
samples precesion is limited.
So while in THEORY the transform is reversible, in practice it is not.

-Alex.