Why flashcards from YouTube
YouTube is now a massive course library: MOOCs, university lectures, technical tutorials, educational channels. The problem is watching a 45-minute video doesn't make you learn. You nod, follow along, and 3 days later it's gone. Without testing or repetition, video learning is very inefficient.
Flashcards fix this by turning the video into an active study tool. You watch once, Diane extracts the concepts, then you review with FSRS-5 spacing. The video becomes a durable investment instead of a forgotten view.
Automatic transcription
Diane first grabs the official transcript if available (creator-uploaded subtitles). Otherwise, AI generates an accurate transcription via speech recognition, in multiple languages. The transcript then drives flashcard generation.
Accuracy is high for main languages (English, French, Spanish, German, Italian, Portuguese), with solid handling of technical terms and proper nouns.
Timestamps preserved
Each generated flashcard carries the YouTube timestamp of its source excerpt. To check what the teacher actually said at that moment, click the timestamp and the video opens at the exact second.
This changes the experience: no more scrubbing through a 60-minute video to find a concept. The card takes you straight to the source passage.
Online courses, MOOCs, conferences
Many fitting channels: university lectures (Khan Academy, MIT OpenCourseWare), conferences (TED, technical talks), educational channels (Veritasium, Crash Course), technical tutorials (programming, design, medicine).
For very long videos (over 2 hours), Diane auto-splits using YouTube's chapter table if it exists, otherwise by detecting thematic shifts.
Use cases
Students: convert a MOOC into a midterm deck, turn a conference into durable learning. Self-learners: study a topic via YouTube without losing it. Pros: absorb a video-distributed internal training, keep content from a professional conference.