THIS type of convolution wont really work with pfft, because it is too short.
mirroring an impulse response required to do each sample of A with each sample of B.

there is a third party deconvolution object, you might use this. (building a mirrored
file is the same process than using it later as filter or reverb.)

-110

p.s. i think its called tconvolve~

i am not sure how to write your own using ifft or pfft object, but i thin i can foresee that it will eat up too much CPU anyway.

what will work as an example is to implement it using buffir~.

but of course that has a certain sample limit as well, and wont do much more than a cheap sounding 2-band EQ or one first reflection for a verb. :)

is there a reason why you want to do your own code in MSP? i think its easier to use soundhack, peak, or matlab.

the download for those HISS tools is still dead … anyone?

-110