fftknn learns to recognize any sound, vowel or pitch you teach it and associate it with a MIDI signal. Later, (in real time) when it hears similar sounds it will trigger the learned MIDI signals.

For example, fftknn can be used to trigger different synths depending on the vowel a singer sings into a mic. It could also be used to trigger different drums or samples based on different similar sounding vocal cues.

requires: python numpy pyjack

Technically speaking: fftknn continuously applies an fft to an incoming signal and use knn to classify it against training samples.