Most people who want flashcards from a PDF are not short on material. They are short on time, and they are staring at forty pages that are somehow both dense and familiar.
This is a walkthrough of how to get from that PDF to a set of cards you can actually review. It covers both ways of doing it, because the honest answer is that the right method depends on the PDF in front of you.
First, check whether your PDF can be converted at all
Open the file. Try to select a sentence with your cursor and copy it.
If the text highlights, your PDF contains selectable text. The words are stored as words, and anything — a script, a tool, you — can read them.
If dragging over a sentence highlights nothing, or highlights a rectangle like a picture, your PDF is a scan or an image export. Inside that file is a photograph of a page. There are no words to extract, only pixels.
This distinction is the single biggest source of disappointment with PDF-to-flashcard tools, so it is worth checking before you spend an evening on it. A scanned PDF is not a hard case. It is a different problem, and it needs OCR first.
Rough guide to what you will find where:
- Lecture notes exported from slides or a word processor — almost always selectable text.
- Textbook chapters from a publisher or course pack — usually selectable, though equations and tables are their own mess.
- Papers from a journal or preprint server — selectable, but heavily interrupted by references, figures, and captions.
- A photo of a friend's notes, or a phone scan of a book — almost never selectable.
- Old photocopied handouts distributed as PDFs — usually scans.
If you are in the image camp, stop here and get an OCR pass first. Everything below assumes real text.
The manual method, and where it breaks
The manual method is what most people do, and it works fine on short material.
- Read a section, not the whole document.
- Close or cover the source.
- Write down the questions the section was answering, from memory.
- Go back and check accuracy against the source. Fix anything you got wrong.
- Write the answers under each question.
The reason to write the questions from memory rather than by copying sentences is that it forces you to notice what you did not retain. Copying a heading into a question box feels productive and teaches you almost nothing, because you never had to retrieve anything.
Where it breaks is volume. On a five-page handout this takes twenty minutes and you are done. On a forty-page textbook chapter it takes hours, and the hours come at the worst time — right before you have anything to review. The card-writing eats the reviewing, and reviewing is the part that changes your score.
The other failure is finishing the cards and never touching them. A deck you built and abandoned is worse than no deck, because it cost you the evening.
The assisted method
An assisted tool does the card-writing pass for you, then hands you something to correct.
- Upload the file. In WakeCards you add the PDF, and it previews the cost before you commit.
- Pick a range you can actually finish. This matters more than people expect. Uploading a whole 400-page book produces a giant deck you will not get through. Upload the chapter you have to know this week.
- Read every card and be ruthless. The generator produces a first pass, not a final deck. Delete the vague ones, split the two-in-one ones, fix any answer that does not match your source.
- Answer the cards with the PDF closed. This is the step people skip. Reading a generated card is not studying; answering it is.
- Mark what you missed, and come back. The deck becomes your loop.
The value here is not that software writes better questions than you. It is that the first pass changes the job from "write eighty cards" to "review and fix eighty cards." One of those jobs gets done on a weeknight.
You still do the thinking. You just do it against a draft instead of a blank page.
What to expect from tables, formulas, and slide decks
Three things reliably come through badly, and knowing that in advance saves you from trusting output you should not trust.
Tables lose their structure. A row that meant "for this input, this output" can flatten into a sentence with no clear subject. Any card built from a table is worth checking against the source before you rely on it.
Formulas and symbols do not always survive. Subscripts, superscripts, and notation can come out garbled or rearranged. If your material is equation-heavy, plan on hand-fixing those cards.
Slide exports are choppy by nature. A deck built from bullet fragments produces cards that read like bullet fragments. It often helps to upload fewer slides and let the cards stay coherent.
None of these are reasons to avoid PDFs. They are reasons to treat generated cards as drafts. This is the same posture as the honest limits on the product page: the tool reads text, it does not understand your subject, and anything that looks odd should be checked against the original.
How many cards is too many
A useful rule: your deck should not be longer than what you can review in one sitting, times seven.
If you can review sixty cards in a session, a deck of a few hundred is fine because you will see it across a week. If you build four thousand cards for one exam, you will fail to finish them, feel behind, and stop opening the app. The deck size that kills study plans is not the size that teaches.
Start smaller than feels complete. You can always add the next chapter. You cannot un-quit.
A realistic first pass
If you have a chapter due and one free evening:
- Check the PDF is selectable text. Fix or OCR it if not.
- Upload one chapter, not the book.
- Spend thirty minutes fixing the generated cards. Delete anything vague.
- Run the deck once with the source closed, answering out loud.
- Write down the five you missed. Those five are tomorrow's session.
That is roughly an hour, it produces a working deck, and it leaves you with something you can continue instead of a project you have to restart.
The part that is not about PDFs
The format was never the hard part. Being able to select text is a technical accident that decides whether a tool can help at all. What actually determines whether you learn the chapter is whether you ever answer questions from memory, with the source closed, more than once.
A PDF is just a very good place to get those questions from, because the material is already written and already in your hands. See how it works for the full five-step loop, or read the FAQ if you want the technical constraints in one place.
Frequently asked questions
Can WakeCards read a scanned PDF?
No. There is no OCR step, so a scanned page is an image with nothing to extract. You need to run OCR first, or retype the section.
What if my PDF has text but the equations come out wrong?
That is expected for symbol-heavy material. Generated cards are a draft; fix equations by hand against the source before you rely on them. If the whole document is equations, cards built from prose may serve you better than cards built from formulas.
How do I know the cost before uploading?
WakeCards previews the cost for PDFs at 1 credit per page, and you choose the range before processing starts. Pick a smaller range if your credits will not cover the whole file.
Is it better to upload one chapter or the whole textbook?
One chapter, almost always. A deck you finish beats a deck you start, and you can add the next chapter once the first one is under control.
Can I paste text instead of uploading a file?
No. Paste is not supported; the file has to be uploaded. Text files and Markdown are supported if you would rather convert the source first.