There is a bug in CEP, wherein if you once record a float type 3 file, and
subsequenly record a normat 16bit file, CEP sets the format tag for the latter
as type 3, making the file unplayable outside CEP itself. The trick is to make a
new recording using one of CEP's other (very unofficial!) 32bit formats , which
sets the tag as 1 again.

Richard Dobson

Michael Gogins wrote:
> 
> >the only benefit i can see is IRCAM *supports* floating point..
> >AIFF doesnt - AIFC does (but is not "official" - Madole/Erbe
> >just "made it so") - and im not sure what the story is behind WAV floats
> >(???)
> 
> I don't know the official story on WAV floats, but I do make my pieces in