pszemraj/vid2cleantxt ? reverse-engineered prompt

Build me a Python tool that can take a folder of speech heavy video files and turn them into clean, readable text.

I want to be able to run it from the command line or import it in Python, point it at an input folder, and get back transcriptions plus a separate metadata file with basic run details. It should handle long videos by splitting audio into chunks first, then transcribing each chunk with a modern speech to text model. After that, clean the text up a bit with spelling correction and sentence spacing so the output is easier to read. If you can also extract a few useful keywords from each transcription and save them in a spreadsheet, that would be great.

Make the output folders and filenames sensible, and include a simple example so I can try it on my own video files right away. If anything needs current docs or model details, look them up online while you build it.