Script that extracts the text from all Videos in a Youtube Playlist. (Used to extract the Questions of the 50-Question-Videos from PietSmiet's Youtube Playlist)
The IDs of all videos that have not yet been processed and their length are queried via the YouTube data API. With puppeteer-cluster screenshots are taken every X seconds of the video using multiple chromium instances. These are processed with jimp to finally extract the text from the screenshots with tesseract.
- Install dependencies with
npm install - YouTube-API Key needs to be stored in an enviromnent variable called
YT_API_KEY - Adjust filepath, filename, sreenshot-interval, language, text frame position/size etc. in constants at the top of
parse.js - Run with
npm run start