Transcripts can be pulled from the YouTube API, check the existing YouTube stats script from the marketplace, this seems to be the matter of querying the right property from YouTube’s servers. There’s really no need to reinvent the wheel, unless your goal is to learn by reimplementing a basic Airtable-YT integration from scratch.
There’s even a convenient console in the sidebar to experiment with before getting to the Airtable integration part.
Basically, look into these docs a bit, try to use their info with that script you found, and see if you can manage to poll the data manually with that combo. Afterward is when you should start thinking about how to automate that action, which is still easier than trying to reimplement that Python repo you found and linked to originally.
If that’s too much, Integromat and similar services offer largely drag-and-drop interfaces that should let you do what you want without so much as looking at code. But that will all depend on how often do you need these transcript “pulls” to happen. If it’s not that often, then start with Integromat right away as you might be able to get by just with the free automation quota.