Transcribin

Finds the slide changes in a lecture video and writes notes with one section per slide.

Sampled frames are shrunk to 240 px tall and compared in pairs. A new section starts when structural similarity and histogram correlation both drop, when an occasional OCR check finds the slide's text changed, or when edge and text-area heuristics flip a frame between slide and not-slide. Each caption goes to the section it starts in.

Paste a YouTube lecture link and Transcribin detects where each slide starts, pulls the captions, and builds a Word document where each slide's section holds what was said while it was up. An uploaded video gets a list of slide-change times.

Next: GoCryptA Go CLI that stamps ownership into a PDF and seals it with AES-256-GCM.