Transcribe audio file with automatic language detection

Transcribe an audio file and return the full transcript with per-utterance metadata. Language is detected automatically for each utterance, so different utterances in the same file may be identified as different languages when code-switching occurs.