LLExtractor watches anime, plays games and reads manga with you. Click any word on screen and it becomes a flashcard — with the native audio and the exact scene you learned it from.
Demo clip coming soon — real gameplay, real mining, no mock-ups.
Load any video — subtitles appear even when none exist, transcribed on-device by Whisper AI. Every line is clickable.
Zero-setup text hooking reads visual novel dialogue live, and a floating dictionary sits over your game. No copy-paste, no second monitor.
Point it at a page. It finds the bubbles, reads them in Japanese order, and translates live while you read.
Every card keeps the full sentence, the audio, and a screenshot of the moment. Spaced repetition is built in — no exporting anywhere.
Watch the show, play the game, read the chapter. LLExtractor runs quietly alongside.
One hotkey turns it into a flashcard — sentence, dictionary entry, native audio, screenshot. Under a second.
Cards resurface right before you'd forget them. You remember the scene, so you remember the word.
Mined 38 words from one episode of Frieren and I actually remember them — because I remember the scene. Textbooks never did that for me.
I hooked a visual novel that no other tool could read and it just pulled the dialogue straight into flashcards. The audio on every card is the killer feature.
I learn Korean from dramas, and being able to click a line and get the sentence, the reading, and the clip in one press is exactly what I always wished existed.
The screen OCR reads games that can't be text-hooked. I'm finally mining vocab from JRPGs I've wanted to play in Japanese for years.
500 days of Duolingo taught me less than two weeks of mining anime with this. It fits the way I already watch stuff.
Furigana, pitch accent, and a screenshot on every card, with spaced repetition built in — I stopped exporting to five different apps. It's all here.
Beta invites go out to the waitlist first, in order.
Questions, feedback, or just want to say hi? Reach out.