i would like to program a sort of FFT-Dronemachine, where the incoming Audio should be freezed, and change really slow in reaction to the following incoming data, the result should sound like a drone which moves really slowly but tries to give a spectral print of the incoming audio.
basically the time-behaviour and the spectral changes of the incoming signal should be smoothed out.
in the beginning i thought, vectral~ would do the job, but the f.e. the slide message seems to affect only the slide behaviour, but attacks (slide ) do not really change, they are still there….?
then i thought, i have to interpolate between successing buffers:
what i tried to do is, write the incoming signal into a buffer1, read this buffer1 out, compare it to a second buffer2, the result of the comparision is written into buffer3. this happens f.e. every 2 frames, in the "off"-frame, i write buffer3 into buffer2. so that in the next cycle i compare the new incoming signal with the result of my comparsion of the previous cycle. and, finally, buffer3 is the one i hear then in the end.
i tried to implement a tapin/out between the index~ and poke~ with the delay of 1 framesize. just feeding back the magnitude comes closest to what i imagine,feeding back the phase information has a rather chaotic effect. also i think i have to put the frameaccume-internal-phasewrapping on, other wise feeding back the phase wont work.
but if i feedback just the magnitude, i don