It would be cool if you could set a timeout to automatically submit the voice-to-text, e.g after 3 seconds of silence.