A transcript made from the sound, not from the captions
Most YouTube transcript tools do one thing: they read the caption track that YouTube already stores for the video and show it to you as text. That is quick, but it fails in exactly the cases where you need a transcript most. Many videos have no captions at all, the uploader can switch them off, and the automatic ones often arrive as a wall of lowercase words with no punctuation and no sentence boundaries.
This page works differently. It takes the audio of the video and transcribes it from scratch with a speech recognition model, the same way an uploaded recording is handled. The result has punctuation, capital letters and paragraphs, and it does not depend on whether captions exist. The trade-off is time: you wait a few minutes instead of a few seconds, and in return you get text you can actually quote.