Okay, let's hit the first task on the list:
This seems straightforward. The obvious way to do this is to write some sort of script that uses the YouTube API to search for all videos, and then to retrieve metadata for those videos, so that we can filter out things like YouTube shorts and videos that might not be relevant based on the title, maybe the description, the duration. And then third is we then have to transcribe all of his videos.
I'll be using n8n as the primary interface unless I have a strong reason to switch to something else. It's just easy to run, it's visual, and it's easy to debug as well.
Retrieving the videos
Official channel: https://www.youtube.com/@eyadqunaibi
There are a bunch of playlists already configured so we may be able to use that directly instead of having to crawl and categorize.
okay we'll just go ahead with the simplest possible approach for now which is simply parsing the playlists that are available and then retrieving all of the videos within those playlists and using that as a structure for the data we're going to extract later.
So the first thing we'll do is to create a new n8n workflow and connect the YouTube node to retrieve all of the playlists:
We'll then retrieve all of the videos within those playlists:
And store the results in a database:
That's it! We've got all of the videos in our database.
Next, we'll work on the transcription.