I would like to implement convolution inside phase vocoder. At the moment I am multiplying amplitudes and adding together phase differences (instead of phases). Is that correct or do I have to again phase wrap the result of phase difference summation?

I get the phase differences in pfft~ by using framedelta~ and phasewrap~. At the output of PV is object frameaccum~.

PS: I would much rather do the convolution with phase differences due to the structure of my patch.

Convolution by summing phase differences instead of phases