Documentation
Shiori（栞, "bookmark"）is a Windows-first desktop application for learning Japanese through reading. These pages document what the app does and how to use it.
What Shiori is
Shiori is built around comprehensible input: the primary activity is reading real Japanese text, and every other feature exists to support that. It is not a flashcard driller — it is a reading companion that happens to teach.
You import books, articles, or any Japanese text (paste, .txt/.md, .html
including Aozora Bunko pages, .epub, .pdf), and the app parses everything to
the morpheme level while preserving sentence and paragraph context. While reading
you click any word to see its dictionary entry, usage register, and a
component-by-component explanation of its conjugated form; one click adds it to
spaced-repetition review. The app tracks what you know, grades every document in
your library by difficulty, and tells you what to read next.
Shiori runs fully offline after first launch. The only network features are optional LLM calls for conversation practice and the online catalog fetch in Sources, which falls back to its local cache.
Where to start
New here? Begin with Getting Started, then Reading — the heart of the app. The sidebar lists every topic:
- Getting Started — first launch, reference-data download, importing your first text, the four word statuses
- Reading — clickable tokens, conjugation-aware selection, furigana modes, the reading clock
- Reviews & SRS — FSRS scheduling, in-context cards, cross-book examples, the mark-known-on-finish sweep
- Dictionary & Kanji — JMdict search, kanji cards with readings, grade, and stroke order
- Online Sources — Aozora Bunko and Japanese Wikisource search and one-click import
- AI & Chat — conversation practice, annotation underlines, level calibration, LLM backends
- Statistics — reading velocity and calendar, comfortable reading level, forecasts, retention
- Data & Interop — Anki export/import, settings transfer, database backup and restore
- Architecture — workspace crates, the data directory, the NLP pipeline