Switch uses a speech recognition engine to generate a rough approximation of the words spoken in an audio file. Choose an engine and profile from the list.