Speech synthesis extension prototyping questions

[ericrosenbaum]

• Should we separate this from speech recognition into its own extension?
• To really do this across platforms, I guess we need to set up a server-side speech synth? e.g. MaryTTS