The idea of starting a project to preserve speech synthesis technology
by means of emulation has been in my head for a while.
The emulator would support cores like the TI TMS series, digitalker
and MSM5205, for example.
It would work by reading a file on the operating system, which would
be considered either input to the IC's input pin, or a PROM which the
IC reads information from, and be outputted by an emulated speaker,
which would pipe to the pc speakers.
This way, one by one, all documented and reverse-engineered speech
IC's can be preserved by having cycle-exact output. Much example
source code exists in the MAME project.
While we can not use it, it can serve as a base to be rewritten.
While I have no coding skills myself, I could oversee and organize the
effort, maintain the CVS and website etc. If you are interested, let
me know.