a large speech database synthesis system
Here's a
synthesised explanation
of how it works
and it isn't perfect yet - listen to
Tetsuko's comments
but we know HOW to predict the prosody, and
why
it isn't right
