[SciPy-User] EPD and fast FFT?

Joseph Anderson j.anderson at hull.ac.uk
Tue Mar 23 11:25:46 EDT 2010


Hello Dag,



> The latest EPD comes with Intel MKL; so that's likely what you want to
> use. It uses the FFTW 3 API, so any code you write would be usable with
> either MKL (costs money) or FFTW (GPL).

Ah... I forgot to mention I'm using a OSX on a G5 PPC. From my reading, MKL
isn't available for this.

> Personally I'd use Cython to start wrap the FFTW 3 API... at least if
> you know some C that shouldn't take long.
> 
> Dag Sverre

Hmm... I'm a bit rough on my C...

I have some code written calling scipy.signal.signaltools.fftconvolve. With
small arrays, this is fine, but I have some larger RIRs (room impulse
responses) that just grind to a halt.

I've figured that I can write an overlap-add routine to step through both my
input signal (doing this already) AND the kernel.

However, my feeling is that a nicely written convolution routine will 'play
nicely' and do that for me.

Was hoping that there was a simple way to do that, by just setting up FFTW
or something else to stand in for scipy.signal.signaltools.fftconvolve.




~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dr Joseph Anderson
Lecturer in Music

School of Arts and New Media
University of Hull, Scarborough Campus,
Scarborough, North Yorkshire, YO11 3AZ, UK

T: +44.(0)1723.362392 T: +44.(0)1723.357370 F: +44.(0)1723.350815
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~


*****************************************************************************************
To view the terms under which this email is distributed, please go to http://www.hull.ac.uk/legal/email_disclaimer.html
*****************************************************************************************



More information about the SciPy-User mailing list