RepetoTo turn a YouTube video into flashcards, you need more than a transcript and a button. A transcript records what was said; a useful deck isolates what is worth retrieving later, turns each idea into a clear question, and removes the filler that made the video natural to watch.
Repeto can handle the mechanical path—paste a YouTube URL, generate cards into a deck, and schedule them with FSRS—but the result improves dramatically when you make three decisions first: what you are learning, what kind of answer belongs in this deck, and which generated cards deserve to survive.
Here is the complete workflow.
1. Choose a video that can become a deck
The best source videos are structured and factual: lectures, tutorials, conference talks, explainers, language lessons, and exam reviews. A video does not need chapters, but it does need claims, steps, terms, or examples that can stand on their own.
Be cautious with opinion pieces and highly visual demonstrations. If the central information exists only in a graph, equation, or gesture, the transcript may miss it. You can still make cards, but add the missing visual context manually or use a screenshot as a separate source.
Before generating anything, write one sentence that defines success: “I want to remember the causes and consequences,” “I need the command syntax,” or “I want the new Spanish vocabulary.” That sentence is the filter you will use during cleanup.
2. Put the URL into the right deck
In Repeto, open the deck where the cards should live and paste the YouTube link into the generation flow. The video transcript becomes the source, and a longer job can continue in the background while you do something else.
The destination deck matters because its generation profile defines the shape of a good card. A language deck can ask for a word, translation, grammatical detail, and example sentence. A programming deck should prefer short questions about behavior, syntax, and tradeoffs. A history deck may ask for causes, chronology, and significance.
Use the deck instructions to make the output specific. Useful constraints include:
- “Keep answers under two sentences.”
- “Create cards only from claims made explicitly in the video.”
- “Prefer why and how questions over names and dates.”
- “Include the code example when the answer depends on syntax.”
The goal is not to make the largest possible deck. It is to make cards that match the way you will use the knowledge.
3. Avoid the transcript trap
Weak transcript-to-flashcard tools often cut the text into chunks and turn headings into questions. That produces cards such as “What does the speaker say about memory?” with a paragraph on the back. The wording may be grammatical, but the card is difficult to retrieve and impossible to grade consistently.
A strong card should make sense after you have forgotten the video. It needs enough context on the front, one target idea, and a short answer.
For example:
Weak: What is discussed at 12:40?
Better: Why does a successful recall after a longer delay strengthen memory more than an immediate repeat?
The timestamp may help you trace a card back to the source, but it cannot be the question.
4. Triage the generated batch
Treat generated cards as proposals. Spend five to ten minutes reviewing the batch before it joins your permanent queue.
Delete cards that repeat the same idea, test an anecdote, depend on missing visual information, or answer a question you do not need. Rewrite vague fronts. Verify numbers, quotations, and technical details against the video when precision matters.
Keep the card count smaller than your first instinct. Every card you approve creates future reviews, so a low-value card keeps charging interest. The daily flashcard guide explains how to set a sustainable new-card budget.
This cleanup is where your one-sentence goal earns its keep. A fascinating side story can be true and still be irrelevant to the deck you meant to build.
5. Review by retrieval, not replay
Once the batch is approved, let the cards leave the video behind. Read the question, produce the answer before revealing it, and grade the result honestly. That is active recall; the FSRS schedule then spaces future retrievals according to how well each answer holds.
If a card repeatedly fails, do not automatically study it more often. Inspect it. The prompt may be ambiguous, the answer may contain two facts, or the idea may require understanding that the video never supplied. Repairing one bad card is cheaper than failing it for months.
For the distinction between retrieval and scheduling, read active recall vs spaced repetition. For the memory model behind the intervals, read how FSRS works.
A faster version for language videos
When the goal is vocabulary, you usually do not want a summary deck. Ask the deck to extract useful words and expressions from the transcript, preserve the sentence where each appeared, and add the translation or grammatical fields your language requires.
Context is the advantage of starting from a video: the words came from material you chose and understood, not from a generic frequency list. Our AI vocabulary flashcard workflow shows how to structure those cards.
From watching to remembering
Watching a clear explanation creates understanding in the moment. It does not schedule a future opportunity to retrieve the idea. Turning the video into a small, edited deck closes that gap:
- Choose a factual video and define the learning goal.
- Paste its URL into a deck with the right generation profile.
- Reject transcript-shaped questions and long answers.
- Keep only cards worth reviewing in the future.
- Let active recall and FSRS do the long-term work.
Paste the next video you are supposed to learn from into Repeto. The transcript becomes the raw material; your judgment turns the generated batch into a deck.