microsoft/markitdown ? reverse-engineered prompt

I need a Python tool that can convert all kinds of different files into clean Markdown, mainly so I can feed the text into an AI.

It should handle the basics like Microsoft Office documents (Word, PowerPoint, Excel), PDFs, and HTML files. But I also want it to be smarter. It should be able to process images and use an AI like GPT-4o to generate descriptions for them. It should also do OCR to extract text from images inside the documents.

Let's also add support for transcribing audio files and even pulling down transcripts from YouTube video links. For bonus points, it could handle entire ZIP archives by converting each file inside.

I'd like to use this both as a simple command-line tool where I can just type convert my_file.pdf and also as a library I can import into my own Python scripts for more complex workflows.