Gianni Pavan wrote:
> In the windows software I developed (see previous message or the page
> http://www.unipv.it/cibra/softw.html) the time resolution one can get in=
> real-time depends on the sampling rate (up to 192 kHz on standard windows=
> boards), on the number of channels (1 or 2), on the chosen fftsize (512 t=
o
> 16384), on the step among consecutive transforms (2 samples up to the
> chosen fftsize) and, of course, on the available processing power. The
> window length (16 samples to fftsize) and shape (I like gaussian and
> kaiser-bessel) is not relevant for computation time. The computation
> efficiency can be slightly modified by setting a proper data transfer
> blocksize which affects the latency of the display.
Are you here talking about your DOS program or the Windows program?
Certainly SEA is more limited.
> Very few programs allows to modify all these parameters to get the desire=
d
> frequency-time resolution in real-time operation.
All too true. Most sonograms have very limited modification so as to
keep the amount of programming down. And also not to overload as often.
I'm interested in experimenting with FFT sizes in excess of the 4096 I
have available. I'm a mac user, but have both PC emulation on my mac and
a PC laptop, so could, with some awkwardness run the programs. I'm not a
PC expert, so doing elaborate mods to the OS is something I'd not try.
Walt
________________________________________________________________________
________________________________________________________________________
|