Set up whisper transcriptions for DLME

[jacobthill]

We will need transcriptions for Golistan content and this would be a feature that would benefit DLME significantly. I am proposing that we add a task to airflow to transcribe audio content and export the transcriptions with the other metadata. Stanford (and other data providers) could then get back transcriptions as a benefit of DLME partnership.