Credits and licences

Dictionary and kanji information

JMdict, KANJIDIC2 and KRADFILE/KRADFILE2 © James William Breen and EDRDG, CC BY-SA 4.0. JMdict is obtained via jmdict-simplified, compacted to JSON with selected senses and loanword sources. Kanji information is selected/merged for offline study. Derived datasets retain CC BY-SA 4.0; unrelated app code does not automatically inherit that licence.

Preserve upstream documentation and contributor notices, including KANJIDIC code/descriptor contributors. Other-language dictionary material may have separate rights.

• EDRDG licence and contributor notices
• JMdict documentation
• KANJIDIC documentation
• jmdict-simplified
• CC BY-SA 4.0

Stroke diagrams

KanjiVG © Ulrich Apel, CC BY-SA 3.0. SVG paths and number positions are extracted into JSON shards for offline animation/writing practice. Adapted stroke datasets retain CC BY-SA 3.0.

• KanjiVG
• CC BY-SA 3.0

Examples and vocabulary

Examples are selected from the Tatoeba/Tanaka Corpus distribution and annotated with readings. The builder identifies this distribution as CC BY 2.0 France; sentence-level author/source provenance still needs verification before commercial release.

Extra vocabulary uses open-anki-jlpt-decks by jamsinclair/contributors (MIT repository licence). Its earlier chyyran/tanos.co.uk deck sources require provenance review; a repository licence does not settle every upstream content right.

• Tatoeba terms
• CC BY 2.0 France
• open-anki-jlpt-decks

Other kanji data

Character lists use kanji-data by David Gouveia (MIT). Rankings use kanji-frequency by Dmitry Shpika (CC BY 4.0). Pitch accents use Kanjium by mifunetoshiro (builder identifies CC BY-SA 4.0; verify exact source distribution). BabelStone IDS is maintained by Andrew West and explicitly permits personal/commercial use. Inputs are selected/transformed for offline study.

• kanji-data
• kanji-frequency
• Kanjium
• BabelStone IDS permissions
• CC BY 4.0

Listening voices

VOICEVOX:冥鳴ひまり. The current manifest lists her in mixed dialogues; her terms permit commercial use with credit. Other recordings use ResembleAI Chatterbox multilingual, including the owner’s Custom A male voice made from a Qwen3-TTS anchor. The current manifest does not list 青山龍星.

Recordings are synthetic speech. Model/code licences do not prove rights to every reference recording or a real person’s cloned voice. Confirm anchor/model provenance before commercial publication; retain required notices and watermarks.

• 冥鳴ひまり terms
• VOICEVOX terms
• Chatterbox
• Qwen3-TTS

Software and fonts

React, Capacitor, Firebase, DOMPurify, ts-fsrs and other packages are credited in the accompanying software notices. Google Fonts families have individual upstream licences. Online APIs are credited as providers, not represented as granting a commercial service agreement.

• Bundled software notices
• Google Fonts

Open data reuse

Bundled data remains subject to the licences above. Dictionary JSON, kanji stroke shards and metadata/search/pitch data are transformed outputs. Keep their attributions/licences with redistribution. Paid features must not claim exclusive rights to those datasets.