Things it would be nice to do :

   Re-write parse - without using f*lex as it is overkill.

   Add front-end "navy rules" from `phon' or `speak' to
   generate phonemes for gen.c.

   Modify rules for English vs American (keep both!)
    - Support the phonemes in the 'text710.dat' dictionary

   GUI for "klatt" and/or parameters

   Try and get data to pipeline right through in one program,
   while still allowing it to be split.

   de-FORTRAN the code some more.




