Pull speech synthesis into its own module to hot-swap on different OSs

[j6k4m8]

Right now we're macOS-specific due to the calls out to say. We should have a speech synthesis module that uses whatever is available, and supports:

• Volume
• Rate
• Pitch