From the original SynthMaker forum (2004–2014), with its members’ names and dates.
mindplay ·
Finally, a modular synthesizer with actual, factual FFT components! 🙂
Or what? ... I mean, can the FFT component actually be used for other than analysis? - could it be used for, say, resynthesis, for example? Or EQ'ing? Will actual FFT effects and synthesizers be possible to do in SM, or are the components for analysis only?
jorge ·
mindplay wrote: can the FFT component actually be used for other than analysis? - could it be used for, say, resynthesis, for example? Or EQ'ing? Will actual FFT effects and synthesizers be possible to do in SM, or are the components for analysis only?
The FFT component currently available is only for displaying FFTs, not for real time processing. We have in our internal version the beginning of real time FFT but it introduces some delay, like 256 if you're doing 256 samples delay etc, that means that if you feedback across an FFT component, you wont have a 1 sample feedback path. I guess its still useful. So I expect you'll have real time FFTs in the near future and some other low-level components to play with it.
lalo ·
jorge wrote: The FFT component currently available is only for displaying FFTs, not for real time processing. We have in our internal version the beginning of real time FFT but it introduces some delay, like 256 if you're doing 256 samples delay etc, that means that if you feedback across an FFT component, you wont have a 1 sample feedback path. I guess its still useful.
yes it is 🙂
i think we have to be prepared about the FFT/iFFT is goin to work on a 'sample frame' basis..the algorythm itself work on a block of samples (a power of 2 number like 128 or 256 or 512)
being able to choose the block size is essential 🙂
love
lalo
toddd_r ·
Found a link for a realtime FFT method, http://139.134.5.123/tiddler2/c3508/reftr.htm?
NothanUmber ·
one more vote for (at least nearly) real time FFT/iFFT here 🙂
Greetings,
NothanUmber
P.S.: would be nice if one could reach a minimal delay of 64 samples but perhaps this triangle interpolation technique mentioned by toddd_r helps to go even below that limit...
Greetings,
NothanUmber
Curator_Ami ·
Delay could perhaps be reduced by overlapping the sample windows and applying some sort of hanning or hamming window based upon the number of sample beetween the windows, if you see what i mean? So that the window gain is at -6dB at the boundaries of window step size
Further, you could duplicate and mirror the sample before applying the window and analysing perhaps? It would probably make a groovy effect, even if its not neccesarily going to reproduce the original sound accurately. I don't know, I haven't tried it yet =)
Specifically, I would like to see some sort of vocoder implemented. Real time FFT would make that much easier, I think.
Wrote something here? To have a post removed or your name hidden, write to [email protected].