From: "SxDx" <sed(a)free.fr>
Maybe that's how they do it in the video. I gave a try to mbrola and the
result is much more realistic than espeak. Mbrola uses recorded
phonemes (if I understand correctly).
just for the record, there is "festival" which seems
to use similar approaches than (to?) mbrola.
http://www.cstr.ed.ac.uk/projects/festival/
It's free software (x11-like license).