[music-dsp] Choice between 32-bit integers and floats ?

[Johan BOULÉ]

Hi,
I'm a newbie to dsp but I've done a softsynth which internally uses streams of 32-bit samples (for each channel). Since I want to
apply in real-time many filters to this stream, I first apply a preamplification to the 16-bit inputs to use the maximum amplitude I
can obtain with a 32-bit precision. After the filters, I reduce the amplitude to convert the data back to 16-bit.
Here is my question. Why does the VST plugins specification require samples made of single precision floats in the range [-1,+1] ?
It appears to me that the only theoric difference between 32-bit integers and "32-bit floats in [-1,+1]" lies on the bounds due to a
different scale. But in practice, floating point operations consume more CPU.
Can someone explains the advantage of the VST specs ?

Johan

______softsynth sources for Delphi____ http://johan.boule.online.fr


dupswapdrop -- the music-dsp mailing list and website: subscription info,
FAQ, source code archive, list archive, book reviews, dsp links
http://shoko.calarts.edu/musicdsp/