Implement the Whisper backend

[akolson]

Overview

To enable automatic speech recognition(ASR) capabilities in Studio, we need to create a Whisper backend class that implements the base Backend class in the appnexus library. The Whisper backend class should be able to make a request to any whisper model hosted anywhere, for example Hugging Face, Open AI, Assembly AI, etc. As a start, it will perform transcription requests for Studio.

Description and outcomes

• Create a new class Whisper that implements Backend:

CLASS Whisper IMPLEMENTS Backend:
	METHOD make_request(params):
		# make request to DeepLearning models hosted as services(e.g via Hugging Face)

	METHOD connect(params):
		# Implement the connect method for hosted service if any

	CLASS METHOD _create_instance(cls)
		# initialize a Whisper Backend instance

• The class should connect to whichever backend the whisper model is being hosted(if required) by implementing connect().
• Implement the make_request() method that performs transcription requests. It should return the result of a transcription

Accessibility requirements

Not applicable

Acceptance criteria

1. The Whisper backend is created and it implements the Backend class.
2. The class connects to whichever service(Hugging Face, Open AI, etc) the Whisper model is being hosted(if required).
3. The make_request method is implemented correctly and performs transcriptions as expected
4. Tests are written to validate correctness of the Whisper backend logic.
5. Documentation has been added to the backend class, explaining its purpose, inputs and outputs.

---

Resources

[akash5100]

I would like to work on this issue.