[PD] FFT question

julien.breval at tremplin-utc.net julien.breval at tremplin-utc.net
Tue Nov 25 03:32:25 CET 2003


do you mean direct resynthesis (using only inverse FFTs, without forward FFTs) ?

a solution could be using [tabreceive~ array1], an objects who outputs 
continuously one block of data from array1 (which should have a size equal to a 

in practice, you will have to use two arrays :
*[tabreceive~ array1] is connected to the left inlet of [rifft~] (real part)
*[tabreceive~ array2] is connected to the right inlet of [rifft~] (imaginary 

the idea is to act on the arrays, in various ways
* using [tabsend~] (sends one block of data to an array)
* using [tabwrite], [tabwrite~], etc
* manual drawings
* etc

I don't know if all this works in practice, though ... it's a bit late now, I 
will try tomorrow ...

one solution that works is implementing a spectral oscillator (a series of 
additive sinusoïds that you can control the amplitude of with a graphic array)

the "16.table.spectrum.pd" patch in the "3.audio.examples" folder does exactly 
the only problem is about the number of sinusoïds (in the above patch, there is 
a fundamental frequency and about 30 partials of this fundamental)
you could avoid this problem using the "poly" object, but I don't know (at all) 
how poly works

Selon Thurstan <thurstan666 at hotmail.com>:

> I have a question about the FFT blocks. Is there a way to input
> spectrum-information in stead of time-based information?
> To put it differently: If you put in a sine wave it will be analised and
> output like a sine wave. The spectrum will contain
> one bin. I would like to insert a pulse that would be analised in the
> window
> and outputted as if it were a bin.
> So when you delay the pulse with a feedbackdelay one will hear some kind of
> grain pattern consisting of little windowed sine components.
> I hope my question makes sense ;)
> Greetz Wilf
> _______________________________________________
> PD-list mailing list
> PD-list at iem.at
> http://iem.at/cgi-bin/mailman/listinfo/pd-list


More information about the Pd-list mailing list