[Flac-dev] IEEE float?

[Josh Coalson]

--- Joshua Haberman <jhaberman at ups.edu> wrote:
> FLAC currently supports 8, 16, and 24 bit samples; are there any
> plans to
> support 32 bit IEEE floats?  Is there any inherent reason why
> floating
> point samples would be harder to compress than integral ones?

there are no plans.  the entropy coding methods are all designed
for integers.  lossless codecs have to store the error exactly
and I don't know of any efficient methods for compressing floating
point values.

flac could possibly normalize the floating point data internally
but then the range becomes a problem.

Josh


__________________________________________________
Do You Yahoo!?
Yahoo! Tax Center - online filing with TurboTax
http://taxes.yahoo.com/