That was a stupid comment I put earlier. I don't even remember writing it.
Sounds like you have everything you should need for speech if audio is working. I have two more suggestions:
1) Are the translator.library and narrator.device both the correct version? Perhaps one is a lower version than required by Say on your installation (assuming Say even checks).
2) Do you have other sound generating programs running at any point before you attempt to use the speech software? Perhaps a program you ran earlier tied up the sound resources and didn't free them. I'm pretty sure speech needs all audio channels to work because it uses one channel to modulate the other. Try booting with no startup-sequence and see if it works.
