I need some help to make special use of the FFT.
I would like to separate my sound in, let’s say, 512 frequency band and be able to transpose each band independently.
I guess I have to use FFT. There’s a lot about filtering and convolution in the examples but nothing about transposition. Is it possible?
Here is a Freq Warp made in Jitter, integrated in phase vocoder. You could re-patch it so it would work for real-time signal input.
The sound sample in the patch in question is in the form of FFT Jitter matrix. And unfortunately you can’t record sound as FFT jitter matrix in that patch – I have that in a separate patch. Hence I am sending you also one already recorded FFT matrix, it should load automatically.
PS: In one week time, I will upload about 40 similar patches to the forum together with my dissertation. Stay tuned.
PS2: In the GRM version of Freq Warp I think there is also some (FFT?) filtering going on beside bin shifting.
and the files…
PS: can’t upload to forum?
thanks! it looks great!
i’ll have a look on it to try to understand (I’m new with all that FFT stuff)
That patch is a bit pain-in-the-ass-ish. First of all you need to know the basics of phase vocoder. Then the Jitter version of the phase vocoder (on this page http://www.jeanfrancoischarles.com/ you will find Jean-Francois’s article called A Tutorial on Spectral Sound Processing Using Max/MSP and Jitter and some accompanying patches – this is a must read). I am also using his frame interpolation method in that patch. Then you have lin. or log. scales for both axes, and log. are pretty confusing. Then there is a 40 FFT frames wide moving submatrix that is beeing processed – that matrix could be wide only 1 frame (it has 40 frames/pixels only for visual effect) etc…
In the dissertation I will post at the end of the week I explained everything in details. And first 80 pages is just FFT… I just need to make few corrections