Sources and licenses
Learn about the sources of KotobaHoshi vocabulary, kanji, examples, illustrations and audio, with attribution and license information for the related projects.
Vocabulary and kanji
KotobaHoshi is based on data from OpenJLPT 0.3.0. We retain the LICENSE and NOTICE for the version used. OpenJLPT data and derivative datasets are handled under CC BY-SA 4.0. This does not mean that all application code is covered by the same license.
Spelling, readings and part-of-speech verification come from EDRDG's JMdict. Kanji readings, meanings and radicals come from KANJIDIC2. Reference JLPT classifications and short English vocabulary glosses come from Jonathan Waller's resources. The levels are study references, not officially published JLPT vocabulary lists.
Examples
The original example sentences and English translations sourced from Tatoeba in the base dataset are licensed under CC BY 2.0 FR. The site also includes example sentences and translations supplemented and edited with AI assistance.
Translation and explanation changes
Japanese explanations for learners, Chinese definitions and translations of example sentences are supplemented, translated and edited with AI assistance. Additions and translations of the original data retain the attribution and license conditions required for the source and derivative data.
AI-assisted content
Additional definitions, translations, example sentence editing and illustrations on this site are created with AI assistance. Audio is produced using speech synthesis. We do not guarantee that this content is fully accurate, complete or suitable for every context. There may be errors in meanings, translations, usage, visual details, pronunciation or intonation. The content is provided for learning reference only. If you have questions, consult reliable dictionaries, textbooks or a teacher.
Illustrations
Word card illustrations are created with AI image generation tools to help learners associate words with images. They may not cover every meaning of a word. The license for OpenJLPT text data does not determine the usage rights for these illustrations.
Audio
Word pronunciations and example sentence readings are synthesized using VOICEVOX. Voice credits: VOICEVOX:四国めたん、VOICEVOX:ずんだもん、VOICEVOX:春日部つむぎ.
Audio use follows the VOICEVOX terms of use and the terms for the corresponding voice libraries, including the Shikoku Metan and Zundamon voice library terms and the Kasukabe Tsumugi usage rules. Text data licenses do not replace the usage conditions for audio and voice libraries.
Review algorithm
Spaced review uses FSRS by Open Spaced Repetition. The server implementation references FSRS-6 in ts-fsrs 5.4.2 and retains the relevant MIT attribution.