Custom Japanese dictionaries for Yomitan, covering places, culture, food, folklore, poetry, names, geography and grammar.
The ZIP files are published as GitHub release assets, not committed to Git. Each archive is ready to import directly into Yomitan. Source and attribution information appears in index.json and, where available, dedicated manifests inside the archive.
- Download one or more ZIP files from the latest release.
- Do not unzip them.
- Open Yomitan's settings, go to Dictionaries, choose Import, and select the ZIP.
- Wait for the import to finish. The illustrated dictionaries can take a while, especially on Android or low-memory devices.
Downloading or syncing a ZIP does not automatically import it into Yomitan.
| Dictionary | What it contains | Lookup entries | Download size | Download |
|---|---|---|---|---|
| Japanese Cities & Wards | 815 Japanese cities and wards with readings, prefectures, populations, maps and attributed lead images | 815 | 125.3 MB | ZIP |
| Illustrated Ogura Hyakunin Isshu | All 100 poems with Japanese text, historical kana, poets, source anthologies, public-domain English translations, notes and historical cards | 900 | 3.7 MB | ZIP |
| Japanese Railway Stations | 11,439 current and historical stations and stops, including readings, lines, operators, prefectures, addresses, dates and coordinates | 12,750 | 2.5 MB | ZIP |
| Japanese Regional Cuisine | 1,015 illustrated regional dishes with readings, prefectures, traditional areas, ingredients, history and Japanese/English descriptions | 1,139 | 156.7 MB | ZIP |
| Japanese Traditional Colours | 227 traditional colour names with readings, romanisation, English meanings, exact RGB/hex values and generated swatches | 228 | 754 KB | ZIP |
| Japanese Yōkai Encyclopedia | 387 yōkai and supernatural phenomena with readings, variants, Japanese descriptions, classifications and 279 open or public-domain illustrations | 715 | 41.2 MB | ZIP |
| 日本語文型バンク (NINJAL 2026.01) | 2,394 Japanese grammar patterns with official explanations, connection rules, levels, categories and examples | 2,394 | 2.7 MB | ZIP |
| Former Provinces of Japan | Former provinces with readings, circuit groupings, modern prefecture equivalents and maps | 76 | 2.5 MB | ZIP |
| Japanese Kanji Phonetic Families | Phonetic components, expected on-readings, confidence, family members, exceptions and diagrams | 750 | 669 KB | ZIP |
| Japan's Lakes | 50 major lakes with readings, area/rank data, municipalities, maps and photographs | 54 | 10.0 MB | ZIP |
| Japanese Proper Nouns Mk II | Place names, family names and personal names with readings, explanations, alternate readings and photographs | 15,006 | 292.1 MB | ZIP |
| Japanese Public Holidays (2026) | National holidays with readings, statutory purposes, background and official 2026 dates | 17 | 9 KB | ZIP |
| Illustrated Shinto Kami & Japanese Mythology | 260 kami and mythological deities with readings, aliases, English overviews, Japanese source text, relationships and illustrations | 2,123 | 3.6 MB | ZIP |
| Japan's METI Traditional Crafts | 244 nationally designated crafts with readings, explanations, regions, categories, designation facts and catalogue photographs | 244 | 6.9 MB | ZIP |
| Japan's 100 Famous Mountains | All 100 mountains with GSI summit records, readings, coordinates, regions and photographs | 100 | 13.2 MB | ZIP |
| Japanese Kamon Encyclopedia | 426 family-crest blazons with translations, readings, motif breakdowns and 748 Edo/Wikimedia illustrations | 426 | 23.5 MB | ZIP |
| National Parks of Japan | All 35 current national parks with official readings, overviews, designation data, prefectures and photographs | 70 | 7.9 MB | ZIP |
| Onomanabi - Japanese Onomatopoeia | 1,515 onomatopoeia entries with readings, categories and native pitch accents; 838 have definitions and bilingual examples, and 677 are explicitly marked as missing definitions | 1,515 | 197 KB | ZIP |
| 文法 | 534 grammar notes with explanations, furigana, 4,630 original examples, 2,670 labelled AI bilingual example pairs and two source images; extra lookup aliases preserve source notation | 1,324 | 1.1 MB | ZIP |
The complete release is 694.4 MB. CATALOGUE.json provides machine-readable titles, revisions, sizes and SHA-256 hashes.
A visual place-name reference for all 815 included cities and wards. It is useful when a novel, game, news article or train announcement drops a municipality name without explaining where it is. The underlying city deck is MIT licensed. Every Wikipedia/Wikimedia image has its creator, source page and licence recorded in wikipedia_image_attribution.json inside the archive. The release copy also bundles the GFDL and Free Art License texts referenced by that manifest.
A lookup-friendly edition of the classical anthology rather than a flat list of poems. It indexes poems through several useful forms, which is why 100 poems produce 900 lookup entries. Japanese Wikisource transcription is CC BY-SA 4.0. William N. Porter's 1909 translation and the historical card images are public domain.
A large proper-noun dictionary built from station_database snapshot v20260731. It covers 8,989 open and 2,450 closed stations and stops. Original station names remain the primary terms, while disambiguated aliases make homonymous stations easier to inspect. The transformed compilation follows the upstream CC BY-SA 4.0 licence.
An illustrated food dictionary based on the Japanese Ministry of Agriculture, Forestry and Fisheries collection うちの郷土料理. It includes regional context, ingredients, history and bilingual descriptions. Text is adapted under Japan's Public Data License 1.0. Images are limited to recipe pages that explicitly provide permission-free downloadable image ZIPs, with providers and source pages recorded inside the archive.
A compact visual reference for traditional colour vocabulary. Each entry includes its exact numeric colour values, a readable swatch and contrast guidance. The data comes from an attributed English Wikipedia revision and is distributed under CC BY-SA 4.0.
A monolingual illustrated reference assembled from Wikidata, Japanese Wikipedia and freely reusable Wikimedia Commons media. It is useful for folklore-heavy games and visual novels where ordinary dictionaries often provide only a one-line gloss. Exact article revisions, creators, media licences and transformations are bundled in the attribution manifests.
A Yomitan build of the National Institute for Japanese Language and Linguistics' 日本語文型データベース. It keeps the official Japanese explanations, connection patterns, levels, categories and examples. Source DOI: 10.15084/0002000610. Source licence: CC BY 4.0.
A compact illustrated guide converted from the collection owner's local Anki deck. It links each former province to its circuit and modern prefecture equivalents. The retained map provenance is not independently documented, so consult the archive's attribution field before further reuse.
A study reference for recognising phonetic components and predicting on-readings. It is indexed both by component and by example kanji. The source deck's authorship and image terms are not separately packaged in the conversion.
A 50-lake illustrated reference with source-deck maps and Wikimedia photographs. image_attribution.json records the photograph creators, source pages and licences. The retained source-deck maps have unresolved rights status, as recorded in LICENSE.txt.
The largest archive in the collection, with 15,006 lookup entries for places and people. It was converted from a locally held Anki deck. The original deck author and the distribution terms for its text and photographs were not identified in the package metadata.
A small calendar and background reference for Japan's named national holidays and additional statutory days off in 2026. The official dates and statutory purposes were checked against the Cabinet Office overview linked in the archive.
An English-first guide assembled from Wikidata, Japanese Wikipedia and Wikimedia media. Wikidata facts are CC0 and Wikipedia-derived text is CC BY-SA 4.0. The archive retains a private-use notice because every linked Wikimedia file page must be checked for its exact reuse requirements before further redistribution.
All 244 nationally designated crafts current on 27 October 2025, with bilingual explanations and catalogue photographs. The archive's own LICENSE.txt warns that the Aoyama Square photographs have no stated reusable licence and are marked private-study only. Publication here does not grant additional rights to those images.
All 100 mountains from Fukada Kyūya's 日本百名山, matched to current GSI summit records and illustrated with Wikimedia photographs. The archive records each source file page. Those pages control the exact creator and file-level licence requirements.
A crest dictionary built from the SakanaAI/Kamon dataset and open Edo/Wikimedia sources. Rights-uncertain mon-white material is excluded. The archive retains CC BY-SA 4.0 and per-file notices in its manifests and licence file.
A bilingual guide to all 35 current national parks. Official facts come from the Ministry of the Environment under Public Data License 1.0. Wikipedia text and Wikimedia photographs retain their respective source-specific terms and attribution records.
Converted from the owner's Onomanabi Anki deck. All 1,515 words and readings are retained, with 1,846 native pitch variants. Definitions and bilingual examples are present for 838 entries; the remaining 677 carry an explicit missing-definition warning. Generated glosses and rule-derived pitches keep their source labels. The deck's verified/dictionary-sourced labels are not independent verification by the converter.
The owner confirmed authorship and authorised publication. No blanket downstream reuse licence has been specified. Anki animations, CSS and device TTS are omitted; textual motion labels remain. SOURCE.json records the source hash, counts and field mapping.
All 534 source notes are preserved, including repeated headwords with different explanations. They contain 4,630 original Japanese examples and 2,670 source AI-generated bilingual example pairs, plus meanings, connection rules, JLPT labels, register, nuance and related grammar. Furigana and both referenced images are retained. Every AI section remains explicitly labelled as generated and not independently verified.
There are 1,324 lookup rows: 534 source rows and 790 additional aliases, covering 1,260 distinct search keys. Aliases use printed alternatives, optional particles and readings, or remove a leading attachment marker. Discontinuous patterns and A/B/N/V placeholders are not guessed or joined into invented Japanese, so some patterns need a manual search using their notation. No readings or meanings were invented.
Source links to 日本語NET are retained. The deck's third-party text/image redistribution terms are not established by this conversion; publication at the collection owner's request does not grant a new reuse licence. The bundled font, Anki templates, CSS, JavaScript, TTS and review history are omitted. SOURCE.json and RIGHTS.txt document these limits.
These are custom dictionary compilations, not claims of ownership over the underlying source material. There is no blanket licence covering every archive. Each ZIP's embedded notices take precedence.
The collection owner explicitly authorised publication of all 19 archives. That authorisation does not replace third-party permission or grant downstream reuse rights. Several archives retain unresolved source-deck provenance, private-study media warnings or source-page-dependent licence requirements. These limits are documented above and inside the affected archives rather than being hidden behind a blanket licence.
If you redistribute an archive or reuse its contents, preserve all attribution files and follow every upstream licence and restriction named inside it.
Every published asset passes a complete ZIP CRC test, metadata and checksum checks, lookup-entry counts, contiguous bank numbering, safe path checks and media-reference checks.
GitHub Actions downloads all 19 published archives, imports each one with Yomitan's official importer, and queries a real term from every imported dictionary. Change checks use a pinned compatibility commit for repeatable results. A monthly check follows Yomitan's current default branch to catch future importer incompatibilities.
After downloading the assets and CHECKSUMS.sha256 into the same directory, verify them with:
shasum -a 256 -c CHECKSUMS.sha256Repository checks can be run locally with:
python3 -m pip install zstandard==0.25.0 # Needed for Python 3.12/3.13; optional on 3.14
python3 -m unittest discover -s tests -v
python3 scripts/validate_release.py --assets-dir ./release-assetsThe Onomanabi converter and synthetic-fixture tests are included. To reproduce an owner-authorised build:
python3 scripts/convert_onomanabi.py Onomanabi_1.apkg onomanabi-yomitan.zip \
--revision 2026.09.08-v1 --source-owner bee-sanOnly supply --source-owner after the named owner confirms ownership and publication permission. Private source decks and Anki review history are not committed.
To reproduce the grammar build from the supplied modern-format deck:
python3 scripts/convert_bunpo.py 文法.apkg bunpo-yomitan.zip \
--revision 2026.09.08-v1 --report conversion-report.jsonThe report is local audit evidence, not an additional release asset.