What's the difference between source-filter convolution and vocoding?
I'm looking at the convolution-workshop patch in the FFT fun section of the Msp examples. Does the process at the core of this patch differ from that at the heart of a vocoder?
2 things: phases (i.e. complex multiplication in convo vs amplitude only in voco) and the number of bands (vocoder rarely goes above 24 and convo rarely below 256)
I hope this (very rough) explanation gives you a hint or 2.
i also would say that you can not really compare fft filter bands with other types of filters (such as biquad), which are far more close to analog filterbanks.
I forgot to say that the 16 to 24 bands of a vocoder are usually spread exponentially (same musical interval per band (ratio of octaves)) and the fft bands are spread linearly (same number of Hz per band)... you can always work around those differences, btw...
Thanks for the replies. And yes, those are a useful hints... they hint that I need to go back to the tutorials to fully understand this fft business!