Credits & Licences
Updated July 28, 2026
Learn Mandarin is built almost entirely from open data and open source. This page names every source, what it is used for, and the licence it is used under — both because those licences require it and because the people who made this work deserve the credit.
Language data
CC-CEDICT
English definitions, pinyin and tone marks come from CC-CEDICT, a community-maintained Chinese–English dictionary, used under CC BY-SA 4.0.
CC BY-SA is a share-alike licence, so the vocabulary data this app ships — the deck files
under /decks/ — is itself made available under
CC BY-SA 4.0. You are free to
copy and adapt it, provided you credit CC-CEDICT and release your version on the same terms.
Tatoeba
Most example sentences and their translations come from
Tatoeba, used under
CC BY 2.0 FR. Sentences are
contributed by Tatoeba's members; each one we ship keeps its original Tatoeba sentence id,
so any sentence can be traced back to its source and contributors at
tatoeba.org/sentences/show/<id>.
Sentences were converted to Simplified characters and had pinyin generated automatically; those changes are ours, not Tatoeba's. A small number of sentences are written by us rather than drawn from Tatoeba.
HSK 3.0 word list
The vocabulary and its ordering follow the HSK 3.0 standard published by the Chinese Ministry of Education — bands 1 to 4, in teaching order. Word lists of this kind are factual standards rather than creative works.
Stroke order
Make Me a Hanzi
Animated stroke order and the stroke-tracing practice use character data from Make Me a Hanzi (via hanzi-writer-data), released under the LGPL and the Arphic Public Licence. Its graphics are derived from Arphic Technology's fonts.
Hanzi Writer
Stroke animation and grading are done by Hanzi Writer by David Chanin, under the MIT licence.
Typeface
AR PL UKai CN
Chinese characters on the flashcards are set in AR PL UKai CN, © 1999 Arphic Technology Co., Ltd., used under the Arphic Public Licence. It is the same typeface the stroke-order outlines were drawn from, which is why a character looks identical before and after you reveal it.
The copy we serve is modified: it was subset to the 1,170 characters this app displays, and converted to WOFF2. No outlines were redrawn. The full modification notice is at /fonts/README.md, and the subset remains available under the Arphic Public Licence.
Audio
Spoken audio for words and sentences was generated with Google Cloud Text-to-Speech
(cmn-CN-Wavenet-A) and is served from this site. It is synthesized speech, not
a recording of a person.
Software
The app is open-source software, used under the MIT licence. The MIT licence requires its copyright notice to travel with the software, so:
| Component | Used for | Copyright |
|---|---|---|
| Preact | User interface | © 2015–present Jason Miller |
| ts-fsrs | Review scheduling (FSRS) | © 2026 Open Spaced Repetition |
| Hanzi Writer | Stroke-order animation | © 2014 David Chanin |
Each is provided "as is", without warranty of any kind, express or implied, including but not limited to the warranties of merchantability, fitness for a particular purpose and non-infringement. In no event shall the authors or copyright holders be liable for any claim, damages or other liability, whether in an action of contract, tort or otherwise, arising from, out of or in connection with the software or the use or other dealings in the software.
Corrections
Definitions and example sentences come from community-maintained sources and can contain mistakes. If something here is wrong, or a credit is missing or incorrect, please tell us: [email protected]