I guess could work, real mic -> stt -> tts -> emit from a virtual audio device configured as the mic...