To continue the OP's question, I'm looking to use Talkie on a Mega and I already saw in the Talkie library page that it doesn't work on a Mega but I'm wondering why and how much work it would take to fix it so it does work? I very much need some sort of voice synthesis on the Mega. Are there alternatives? Thanks.
So far in my searching, I've been unable to find any support on the Mega, which is kind of disappointing. First I find out the comparator doesn't work because it wasn't exposed on the board and now it seems the Mega is sufficiently different in architecture that many libraries don't work on it. This makes design decisions annoying - I chose the mega for its abundance of I/O, particularly the analog inputs, but I can't do voice on it so now I have to use I2C to command voice on a slave like a nano or something.