Florian is a prototype software + hardware text-to-speech solution. It provides a command-line tool to take program arguments as input, uses the CMU pronouncing dictionary to first convert text to phonemes, then to Speakjet phoneme codes.

It then uses the built-in Teensy UART functionality to stream this to the Teensy over USB, which contains a simple program to interact with the Speakjet and provide buffer management.

The audio output is then run through a passive two pole low-pass filter and an LM386-based amplifier.

CODE

Both the client side code and embedded code may be downloaded here. My c/c++ is a bit rusty, but it should be useful as example code or as as starting point for your own Teensy or Speakjet project.