i'm interested in fourier transformation and how the relative pd objects work (fft~, ifft~, rfft~, rifft~) so i tried to find something in the pd help browser (I01. fourier transformation), in the help patches of the objects and even on wikipedia and the sad truth is:

i don't understand a word!

all i know so far is (at least i do believe), that fourier transformation is an algorithm, use to analyse complex waveforms into their sine components.

now my question: is there some source in the web, where i can get some information on this? suitable for dumbasses like me? maybe even in german language?

i'd be thankful for any hint!

My favorites:

http://www.pd-tutorial.com/english/ch03s08.html
(from http://www.pd-tutorial.com)

and

http://www.parasitaere-kapazitaeten.net/Pd/fft_und_pd_en.htm
(found on http://puredata.info/docs/tutorials)

a fourier transform takes a signal in the time domain (amplitudes over time) and transforms that same signal to the space or frequency domain (amplitudes over frequency) and vice verse.

fft~ and ifft~ are so called because it's the fast fourier transform, a specific algorithm that does the job relatively quickly.

http://en.wikipedia.org/wiki/Fast_Fourier_transform

As you might've guessed fft~ transforms to the frequency domain while ifft~ transforms it the time domain. In PD when you do an ifft~ you have to normalize the signal. I believe it's divide by the blocksize (64).

fft~ outputs two signal, the first are the frequencies and the second contain those frequencies' phases. The number of pairs it outputs is related to the blocksize. So at the default fft~ will only get you the first 64 frequencies present in the original signal.

I have very little experience using FT and doubly so for doing analysis so I can't tell you more than this. In synthesis FT is important for convolution. Convolving in the time domain is more expensive computationally than doing it in the frequency domain and that includes the FFT's you would need to transform them there and back. It's also used in time stretching / pitch-shifting.

thank you both!
@ kyro: i don't know how often i've been on pd-tutorial.com, but didn't even think about it.... and there's also a german version! thank you!

its good to have grounded understanding of FFT, even the basics so that you atleast know what is happening, but any deeper than that for me its a complete headfuk.

THat is why I generally end up using FFTease... cop out I know

you're welcome, toxonic.

boonier, i think i'll follow you path. i did'nt kow about fftease and several ideas i had have already been implemented there. thanks.

Didn't know about FFTease. Thanks for the tip.