* The functionality of speechd should actually be implemented
*in* a speech synthesis program.  I've sent emails to the
creators of festival incouraging them to do so, but I have
recieved no responc, and I believe it has not yet been done.  
Perhaps at some point I will dig into the source and implement
it myself.

* Since speechd was first created, I believe new free speech
synthesis packages for Linux have been released.  Support must
be added.

* finish reformatting CHANGELOG

* fix duplicate connections to Festival when re-establishing a broken
  connection (doesn't appear to hurt anything, but doesn't look right)

* break on sentences instead of newlines
  p71 of the Camel book

* wrap while (<FIFO>) in while (1) {open;;close}

* document the fact that /dev/speech should be oppened for appending,
  never for over-writing.

* install for using in a remote shell connected to a local(?) festival
  server.