Controlling Samples by Spitting Them

Last night, Nate and I were able to sucessfully and intuitively control up to 3 or 4 individual drums based on different sounds vocalized into a mic. The bass drum would play when 'oooh' was sung, a snare when 'eee' was sung and a cymbal when 'aaah' was sung.  There are still some kinks in the system, but as a proof of concept, it works fairly well.  The mapping between input sound and MIDI event are learned in real time so you are not restricted to different vowel sounds.  The 3 distinguishing sounds could have just as easily been a clap, growl and whistle.  The code: fftknn.