you can create a vocal synth from bandpasses using the biquad
object, thats true, but i dont know if that is what the guy did.
typically you would play a pulsetrain of the basefrequency into
the filters which function as resonators to give the pulse a
certain formant (which would be independent from the frequency,
typically it is just static.)
what you need to have for it is a list of coefficients for
certain vocals. typically you would set up 5 filters, 3 seems a bit less.
the whole thing is a very raw reproduction of the human voice,
but it still works because a human can easily identify anything
human voice/vocals.
setting up 5 bandpasses to build a violoncello formant
synthesizer will definetly not work but with aah ooh uuh we
are getting quite close.
if you search the net hard enough you might find the vocal
coefficients from CHANT. they are for [resonators~] i believe,
with biquad it is more fun (more flexibilty) but a also a bit
more difficult because you have to take care about the gain
coefficient yourself (which is linked to the Q in a biquad).
ich hoffe ich konnte alle klarheiten beseitigen.
-110
.
