converting data to sound and back.

This may be off the wall but, could integers emitted as tones of a
pitch and duration have a digital record that would be smaller than
the source data or would it be larger?

I'm thinking of Two states ( tones ) only.

Heh.. switched to Decaf.. Maybe I should switch back :)


Ernst

Ernst_Berg@sbcglobal.net (Ernst Berg) wrote in message news:<be9ae35b.0409300746.6be5ef92@p
osting.google.com>...
> This may be off the wall but, could integers emitted as tones of a
> pitch and duration have a digital record that would be smaller than
> the source data or would it be larger?

Shannon's noisy channel capacity theorem says the best you can do is B
log2(1 + S/N) bits per second, where B is the bandwidth in hertz, and
S/N is the signal to noise ratio.  It doesn't matter what code or
representation you use.  Any compression would have to be achieved by
modeling the probability distribution of your source.

-- Matt Mahoney

matmahoney@yahoo.com (Matt Mahoney) wrote in message news:<8a1ed69a.0410010915.10b4497e@pos
ting.google.com>...
> Ernst_Berg@sbcglobal.net (Ernst Berg) wrote in message news:<be9ae35b.0409
300746.6be5ef92@posting.google.com>... 
>
> Shannon's noisy channel capacity theorem says the best you can do is B
> log2(1 + S/N) bits per second, where B is the bandwidth in hertz, and
> S/N is the signal to noise ratio.  It doesn't matter what code or
> representation you use.  Any compression would have to be achieved by
> modeling the probability distribution of your source.
>
> -- Matt Mahoney

Thanks Matt

I'm learning.

Ernst