So I've looked into it a little more and from what I gather from the thread about controls, it seems like the planned implementation wouldn't require you to sing the note perfectly, but would instead pick the note nearest to your current pitch. Maybe it was scrapped due to complexity, maybe due to accessability concerns, I'm not sure, either way still really cool. Wish I could find more information, but glad to have found a little more of the tale. I'm glad the threads from 2007 are still around ~