Nice work! Do you think this work could be adapted to recognize human-sung notes in real time (e.g. turn my humming into an orchestral melody)? If so, let's get in touch!
It can absolutely do near-real-time pitch detection of singing or humming or even better whistling. We may want to tweak the parameters a little bit for that specific domain but it probably does reasonably well right now.
If you want you can test out how well it does as-is right now. Within the app go to "Settings" then "Audio Analysis Settings" then click the "Test Settings" button. That will show you the same pitch and note-onset detection as used during game-play.
I'd be happy to talk more -- you can reach me as rod@ the domain I linked to -- but to be candid monochromatic pitch detection of the human voice is not a hard problem. I'm not sure FATpick is as special at this as you seem to think, but I'm not quite sure what you have in mind. And I certainly don't want to talk you out of thinking that :) I'd love to hear more about it.