A child’s voice is broken down into individual frequencies that are then turned into MIDI events used to mechanically control a piano. For real!
http://www.todayandtomorrow.net/2009/10/08/speaking-piano/
It’s interesting to hear the composers talk about a bridge between language and music happening when subtitles help understand the text. I wonder if their goal was to create a half-comprehensible speech synthesiser, or if they stumbled upon this explanation after failing.
They also don’t have a word to describe atomic sound events, so they use “pixels” to refer both to pixels on the screen describing a MIDI note, and the sound it produces as a discreet part of a larger phoneme.