jpff@maths.bath.ac.uk wrote:

> Message written at 04 Aug 1998 09:11:15 -0400
>
> is anyone using IRCAM format, or is everyone using AIFF or WAV?  If no
> one is using it we can remove the code, which would be one less thing
> to maintain.  If it is in use please let me know.
> ==John ffitch


Many apps that originated on NeXT (mxv, for instance) handle ircam
format, as well as the NeXT/Ircam hybrid format (common at least among
NeXT apps ;-) ) The point Matt Ingalls makes about floating-point file
formats should be taken to heart: almost all my work is in floating
point, until the final mix.
This reduces distortions due to quantization error: unless you are some
sort of miracle recording deity, you probably don't get samples to
inhabit all 16-bits of resolution: if then you work with these samples
in integer,doing mixing, processing, etc., the distortion just
*multiplies*: a "gritty" sound emerges from
what you thought were clean source sounds! The answer: work as much as
possible in floating point:
once you have converted to float, and normalized the sound, processing
(which *should* be done in floats)doesn't introduce any more distortion
than was present in the original recording. This is not theoretical
la-di-da: I have seen it again and again: students I've had were ready to
toss their projects in the trash until I got them to work in
floating-point. I'm not a teacher anymore (sigh), but perhaps someone
will try this out and learn.
Why the heck isn't there *more* emphasis on floating point in musical
signal processing ? Makes all the difference

Char lieB


--
*********************************************
Charlie Baker              baker@charlieb.com
"Das Ewig-Weibliche Zieht uns hinan." -Goethe
*********************************************