Settings

Settings are one scrollable page with two sub-pages: Dictionaries and Anki. The back button or gesture returns to the list.

Language

Language — the language you study: Japanese (default) or English. Lookups, the Dictionaries page and OCR clean-up follow it. English also lists its optional English OCR models (23 MB, PaddleOCR): Download, a progress bar, Installed · Remove, or Retry after a failure. See Languages.

Dictionaries — install Jitendex, JMdict, JMnedict, Pixiv Light or Kanjium pitch accents with one tap, import Yomitan zips from files or Google Drive (your own Drive or a shared file/folder link), drag cards to reorder priority, check for updates, inspect entries, delete. Name dictionaries such as JMnedict start at the bottom so they never shadow vocabulary.
Anki — backend (AnkiDroid or AnkiConnect), URL, deck, note type, tags, whether to attach the screenshot, the picture’s maximum height (default 462 px, 0 = original), a Test connection button, and a per-note-type Fields editor that overrides name matching.

Engine — meikiocr (recommended after the model download) or Google ML Kit. The explanation under each option says whether the model is installed.
Download meikiocr models (31 MB) — shown when meikiocr is selected but not installed. Until the download finishes, ML Kit is used.
Two-pass OCR (ML Kit) — re-read detected lines at higher resolution and keep the better result. More accurate, a little slower.
Filter non-Japanese lines — drop UI noise such as menus or numbers when Japanese is present.
Live text window — show the floating feed of new lines while auto OCR runs.
Tap text on screen to look it up — the invisible screen dictionary layer.
Pause the video while looking up — media-key pause/resume around a lookup.
Auto OCR interval — 200–2000 ms (default 400 ms). Lower catches short lines sooner; higher saves battery.

Glossary — what the card’s definition holds, like Yomitan’s markers:
- First dictionary only (default) — only your highest-priority dictionary ({glossary-first}), the one at the top of Settings → Dictionaries.
- All dictionaries, unnamed — every dictionary’s definitions without the dictionary names ({glossary-no-dictionary}).
- All dictionaries, named — each dictionary under its name ({glossary}).
In the lookup popup, tap a dictionary’s name (Use on card) to put only that dictionary on that word’s card instead.

Capture game audio — playback capture through MediaProjection. Requires the microphone permission and Android 10+.
Microphone fallback — record the microphone when device audio is silent (DroidDeck, Netflix). Works with the speaker on; records nothing with headphones.
Noise suppression — Android’s voice noise filter on the mic session.
Speaker correction EQ — high-pass plus a per-device curve measured from the tablet’s own speaker, so mic clips are less tinny. Includes a Measure action.
Compressor — evens out loud peaks before normalization.
Normalize volume — bring every clip to the same loudness.
Word audio — fetch pronunciation from JapanesePod101/Jisho.
Custom audio URL (optional) — tried first; {term} and {reading} are replaced.
Replay buffer — 1–10 minutes (5 by default, about 2.9 MB of memory per minute).
Start offset / End offset — 0–3000 ms of audio kept around the line.
Trim silence (experimental) — VAD trim. Off by default because loud music can clip speech.

The About card shows the app version and credits. The full list is in Credits.

Check for updates — asks GitHub for the newest release; if it is newer, Update to vX.Y.Z downloads it and opens Android’s installer.
Check for updates automatically (default on) — the same check each time the app opens. The update itself always waits for your tap.