using Win SpVoice Interface (SAPI) with python to generate audio files with either text-to-speech or phoneme-to-speech