Choose an approach for producing speech-to-text output, given media file input (let's call this speech_to_text_generation_service for now?)

[jmartin-sul]

possible approaches:

• [investigate/prototype] speech_to_text_generation_service approach 1: Define a Docker container for running open source Whisper in a container that we define and for which we manage deployment (lives in this repo?) #3
• [investigate/prototype] speech_to_text_generation_service approach 2: Explore AWS SageMaker #4

investigate these options, and pick one of these; or add/vet/choose an option we haven't yet thought of.