Extract Chinese Vocabulary Into Flashcards With AI

Paste any Chinese article, message, or paragraph into Flashcard, and AI extracts the 2-3 character words worth learning — each with pinyin, a definition, and an example sentence — then turns them into a deck you review with a Reveal/Pass/Again spaced-repetition flow and can export as a shareable URL, CSV, or clipboard text. No pre-made deck, no manual typing, no install.

I built Flashcard because I kept running into the same problem learning Chinese: I’d find a genuinely interesting article — a news story, a WeChat post a friend sent me, a paragraph from a novel — and I’d want to actually learn the vocabulary in it, not the vocabulary in whatever generic HSK deck I’d downloaded months ago. Pre-made decks are fine for building a foundation, but they’re disconnected from what you’re actually reading right now. The words you need are sitting right there in the text in front of you. Extracting them by hand — looking each one up, writing pinyin, finding an example — is exactly the kind of repetitive task that AI is good at automating, so I built a tool that does it in one paste.

This guide walks through exactly how that works: how to extract Chinese vocabulary from an article and get it into flashcards in 2026, with a real worked example so you can see the actual output before you try it yourself.

How do I extract vocabulary from a Chinese article?

The workflow has three steps, and none of them involve manual entry:

1. Paste the text. Open flashcard.fulinlabs.com and drop in any Chinese text — a full article, a single paragraph, a text message, a subtitle line, whatever you’re currently reading. There’s no minimum or maximum length requirement baked into the workflow; you can extract from a sentence or a whole essay.
2. Let the AI extract the vocabulary. Flashcard sends the text through an AI model (via OpenRouter) that identifies the 2-3 character words in the passage, then generates pinyin, a definition, and an example sentence for each one. This is the part that used to take me twenty minutes per article with a dictionary open in another tab.
3. Review the deck. The extracted words become flashcards immediately. You study them with the Reveal, Pass, Again review system, which is described in more detail below, and covered in depth in my companion post on the spaced repetition method behind Flashcard.

The whole thing runs in the browser. There’s nothing to install, and the app is free to use.

Can AI make flashcards from my own text instead of a pre-made deck?

Yes — and this is the core difference between Flashcard and a typical flashcard app. Most flashcard tools start from the deck: you pick an HSK level or a pre-built list, and you study whatever words someone else decided belonged in it. Flashcard starts from your text. You bring the article, the AI brings the extraction.

This matters more than it might sound like at first, for a few reasons:

• Relevance. If you’re reading a specific article, the vocabulary in it is, by definition, vocabulary you need right now — not vocabulary someone else guessed you’d need at your level.
• Context. Because the words come from a real sentence you were reading, the example sentence generated for each card is generated with that same real usage in mind, not a generic textbook sentence.
• No curation bottleneck. You don’t wait for someone to publish a deck for the topic you care about — sports, tech news, a specific author’s writing style, whatever it is. If you can find the text, you can turn it into a deck.
• It compounds. Every article you read becomes a deck. Over time you build a personal vocabulary library that tracks exactly what you’ve actually been exposed to, which is a much more honest record of your reading history than a stock deck ever was. I wrote more about why native content works better than static wordlists in this piece on learning vocabulary from native Chinese content.

The tradeoff, to be fair, is that AI extraction is not infallible — occasionally a definition or pinyin reading is slightly off, especially for words that are ambiguous out of context or names that get misidentified as vocabulary. I’d treat it the same way you’d treat a good but not perfect dictionary lookup: fast and right the overwhelming majority of the time, worth a second glance if something looks unusual.

A worked example: from sentence to flashcards

Let’s actually do this, rather than just describe it. Here’s a short, ordinary Chinese sentence — nothing special, the kind of thing you’d encounter in a news summary or a casual chat:

这家公司最近宣布了一个新的环保计划，希望能减少浪费。

(“This company recently announced a new environmental protection plan, hoping to reduce waste.”)

Paste this into Flashcard, and the extraction step pulls out the 2-3 character words worth learning, including:

Word	Pinyin	Definition
公司	gōngsī	company
最近	zuìjìn	recently
宣布	xuānbù	to announce
环保	huánbǎo	environmental protection
计划	jìhuà	plan
减少	jiǎnshǎo	to reduce
浪费	làngfèi	waste; to waste

Each of these would come through as its own flashcard with an example sentence — for instance, the card for 宣布 (xuānbù) might carry an example like 政府宣布了新政策 (“the government announced a new policy”), giving you a second real usage beyond the original sentence.

Notice what didn’t get pulled out: single characters like 这 (this) or 了 (a grammatical particle), and function words that aren’t really “vocabulary” in the sense you’d want to drill. That’s the extraction doing its job — surfacing the words that are actually worth spending review time on, not padding the deck with everything in the passage.

From there, you’d move straight into review. Each card shows you the Chinese word first. You try to recall the pinyin and meaning, then:

• Reveal — flip the card to see the pinyin, definition, and example sentence, plus hear it spoken aloud via the built-in text-to-speech audio.
• Pass — mark that you knew it, which pushes the card further out on its review schedule.
• Again — mark that you didn’t, which brings the card back sooner so you see it again before you’d have forgotten it.

That Reveal/Pass/Again loop is the entire review mechanic, and it’s what drives the spaced-repetition scheduling underneath — a topic worth its own explanation, which is why I wrote a separate breakdown of how the SRS method works in Flashcard.

How does the Reveal/Pass/Again review system work?

At its simplest, this is a three-button loop that decides how often you see a given card again:

• You look at a card and try to recall it before revealing the answer.
• Tapping Reveal shows the pinyin, definition, example sentence, and gives you audio playback of the correct pronunciation.
• Tapping Pass tells the system you knew the word, and it schedules that card further into the future — you won’t see it again until you’re close to the point of forgetting it.
• Tapping Again tells the system you didn’t know it, and the card comes back into rotation sooner, so the gap between exposures stays short until it sticks.

The effect over a study session is that your review time naturally concentrates on the words you’re actually struggling with, instead of splitting evenly across a deck where half the cards are things you already know. If you extract vocabulary from five articles over a week, the cards you keep marking Again are the ones that keep resurfacing, while the ones you Pass repeatedly fade into longer and longer intervals. That’s the mechanism, and it’s the same one used across the app regardless of which deck — or which language — a card came from.

Why not just use a dictionary app instead?

I get asked this a lot, since it’s the obvious alternative: why not just tap each word in Pleco or a browser dictionary extension as you read? The honest answer is that a lookup tool and a flashcard tool solve two different problems, and most people learning Chinese only have one of them covered.

A dictionary lookup answers “what does this word mean, right now, in this sentence.” That’s useful in the moment, but it doesn’t leave anything behind. You look it up, you understand the sentence, you move on — and unless you separately copy the word somewhere and build a review habit around it, you’ll be looking the exact same word up again in three weeks. I did this for a long time before building Flashcard, and I’d guess I looked up 报道 (bàodào, “report/news story”) a dozen separate times across different articles before it actually stuck, purely because nothing was tracking that I kept failing to recall it.

Flashcard is built around the second half of that problem: it assumes you can already find the meaning of a word (that’s the easy part, and the AI extraction handles it automatically), and it puts its effort into making sure the word actually sticks after you’ve seen it once — which is the part a plain dictionary lookup never touches. That’s the entire reason the extraction step feeds directly into a Reveal/Pass/Again review queue instead of just showing you a translated word list: a list you read once and close is not meaningfully different from a dictionary lookup you forget about. A queue that keeps resurfacing the words you get wrong is what actually turns “I looked this up” into “I know this now.”

If you’re the kind of learner who reads a lot of native content — articles, chats, subtitles — this compounds fast. Every text you paste in adds to one growing, personal, review-scheduled vocabulary set, rather than a scattering of one-off lookups you’ll never see again.

If Anki is the tool you’d otherwise reach for, it’s worth knowing where it fits and where it doesn’t: I’ve written a full Flashcard vs. Anki comparison on card mechanics and setup, plus a broader look at the best Anki alternatives for Chinese reading if you’re weighing more than just these two.

A quick checklist before you start

• Have a piece of Chinese text ready — an article, a paragraph, a message, anything real that you actually want to understand.
• Go to flashcard.fulinlabs.com and paste it in.
• Let the AI extraction run — it pulls 2-3 character words with pinyin, definitions, and examples.
• Review the extracted list briefly; drop anything that’s a name or an artifact of the extraction rather than real vocabulary.
• Study with Reveal, Pass, Again — don’t try to review the whole deck evenly, trust the scheduling.
• Use the text-to-speech audio on new or difficult words so you’re learning pronunciation, not just characters.
• Export the deck via shareable URL, CSV, or clipboard if you want to keep it, share it, or move it into another tool.
• Switch on dark mode or a custom flashcard background if you’re studying at night or just want the app to feel like yours.
• Studying specifically for HSK? See my roundup of free HSK vocabulary tools with pinyin, audio, and export for how Flashcard’s extraction workflow stacks up against structured HSK decks.

The whole point of building it this way was to remove the friction between “I found something worth reading” and “I’ve actually learned the words in it.” Read more about the project on its hub page, or just go paste something in at flashcard.fulinlabs.com — it’s free, and it runs entirely in your browser.