AnkiWeb addon 1324989483
Recognizes hiragana, katakana, and kanji drawn on an in-Anki canvas, inserts them into the active field, and checks kanji against a target with stroke corrections.
Open on AnkiWeb GitHub Find alternatives
active
| Min Anki | Max Anki | Updated |
|---|---|---|
| 25.08 | 26.05+ | 2026-07-12 |
Loadingβ¦
Bulk-inserts color-coded SVG kanji and optional kana stroke order diagrams into a chosen field, with deck, note type, tag, card-state filters and dark mode support.
Automatically fills Anki fields with kanji readings, hiragana, word type, and limited JMdict definitions offline, plus example sentences, with configurable field mappings and bulk deck updates.
Randomizes Japanese card fonts and optionally swaps hiragana/katakana, displays vertical text, and changes text size during review.
Auto-generates Japanese vocabulary notecards from a word list via the Jisho.org API, including definitions, readings, JLPT level, kanji explanations, furigana and example sentences, placed in a custom deck.
Downloads and attaches audio in bulk from prioritized URL sources to selected Anki cards, with field mapping, kana filtering, request delays, and duplicate-to-empty-field options.
Generates furigana readings for Japanese text, provides Japanese note types, kanji statistics, and dictionary lookup, with customizable furigana display in card templates.
An Anki add-on for practicing Japanese handwriting. Draw a hiragana, katakana, or kanji on a canvas inside Anki, and the add-on recognizes what you wrote and inserts it into the active field. For kanji practice, it can also check your drawing against a target character and show you both what went wrong and how it should look.
β οΈ Early development: Handwriting recognition and writing correction may not yet be accurate enough for reliable use.
Handwriting recognition
Kanji handwriting feedback
The add-on currently uses a recognition engine based on DTW. For performance reasons, the engine is only able to recognize a character correctly when the correct number of strokes were drawn. The following list shows the recognition results using the HANDS-nakayosi_t-98-09 dataset, restricted to samples where the correct number of strokes were drawn.
Sure. Here's the same information as a Markdown list, which is more broadly supported by Markdown readers: Sure. Here's the same information as a Markdown list, which is more broadly supported by Markdown readers:
This benchmark tracks the recognition engines and models the add-on uses, or has considered using over time. The list compares the percentage of correctly recognized characters across all samples from the HANDS-nakayosi_t-98-09 dataset for the DTW engine and Zinnia (using a model from Tegaki).
Correct (DTW)
Correct (Zinnia + Tegaki)
The KanjiVG-derived data in data/ is licensed under
CC BY-SA 3.0. See data/LICENSE.txt for full attribution.
This project uses TUAT Nakagawa Lab. HANDS-nakayosi_t-98-09 stroke database.
An Anki add-on for practicing Japanese handwriting. Draw a hiragana, katakana, or kanji on a canvas inside Anki, and the add-on recognizes what you wrote and inserts it into the active field. For kanji practice, it can also check your drawing against a target character and show you both what went wrong and how it should look.
β οΈ Early development: Handwriting recognition and writing correction may not yet be accurate enough for reliable use.
Try the live demo of the recognition and analysis engine right in your browser. Note that the demo always reflects the latest code, so it can be ahead of what's in the installable add-on.
Handwriting recognition
Kanji handwriting feedback
The add-on currently uses a recognition engine based on HMM.
This benchmark tracks the recognition engines and models the add-on uses, or has considered using over time. The table compares the percentage of correctly recognized characters across samples from 6 writers of the HANDS-nakayosi_t-98-09 dataset for the DTW engine, Zinnia (using a model from Tegaki) and a custom HMM recognizer. The selected writers are evenly split in men and women.
| Hiragana | Katakana | Kanji | Japanese | |
|---|---|---|---|---|
| correct (HMM) | 92.85% | 81.61% | 96.55% | 93.71% |
| correct (DTW) | 59.5% | 62.67% | 54.23% | 53.15% |
| correct (Zinnia+Tegaki) | 21.53% | 2.01% | 47.77% | 45.13% |
The stroke correspondence algorithm has limited support for:
The table compares the percentage of character samples with correctly matched strokes from 6 writers from the HANDS-nakayosi_t-98-09 dataset for each ζΌ’ζ€ level.
| Level | οΌοΌη΄ | οΌη΄ |
|---|---|---|
| Tested | 80/80 | 0/160 |
| Correct | 91.95% | - |
The KanjiVG-derived data in data/ is licensed under
CC BY-SA 3.0. See data/LICENSE.txt for full attribution.
This project uses TUAT Nakagawa Lab. HANDS-nakayosi_t-98-09 stroke database.