Community-oriented audio dataset profile creation

[rgreenberg1]

Description
We will want to add support for audio datasets into GuideLLM to enable the benchmarking of audio multi-modal models like whisper. Since Audio datasets are fairly limited, we should aim to structure the data by use cases in a way where developers can easily understand the context of the data and what is being benchmarked by the model.

User Story
As a developer, I want to benchmark a whisper model so that I can understand performance before I move to production with different audio dataset profiles to make sure my use case (call-center summarization, translation, etc. ) can be met on my target hardware.

Acceptance Criteria

• Enable support for the leading hugging face supported audio datasets: https://huggingface.co/blog/audio-datasets#a-tour-of-audio-datasets-on-the-hub
• Create profiles for the datasets for different use cases in structured folders for the use cases
• • Miltilingual Language Translation
• • English Speech Recognition
• • Speech Translation
• • Audio Classification
• • TBD