Known limitations
This page lists the current boundaries of Origa. Limits are not bugs — they describe what the app does not do today, so you can decide whether it fits your workflow.
Audio transcription accepts WAV only
The speech recognition feature accepts .wav files. Other common formats — MP3, M4A, OGG — are not currently supported. Convert your audio before uploading. See Capture for the full flow.
OCR is approximate on difficult input
Optical character recognition works well on printed text and clean screenshots. It struggles with:
- Heavily stylized or decorative fonts
- Low-contrast or backlit photos
- Vertical text in unusual layouts
- Very small characters
- Handwritten text (not supported)
Always review the recognized words before adding them as cards.
No guest mode
You must sign in to use Origa. An account is required because progress is synced across devices. There is no offline-only or anonymous mode.
First run requires internet
The first time you sign in, Origa downloads dictionaries and content. The first time you use OCR or speech recognition, the corresponding model downloads. After these one-time downloads, the app runs offline. See Getting started for the full breakdown.
Sync requires internet
Your progress is stored locally and synced to the server. If you are offline, progress is queued and synced when you reconnect. You cannot push progress to another device without an internet connection.
Writing practice is animation-based
The kanji writing feature shows the correct stroke order as an animation you follow along with. It does not currently recognize free-form handwriting.
Two-button rating
The rating after each card is binary: Don't know or Know in a regular review, Remember or Don't remember in acquaintance training. There are no intermediate options (no "hard" or "easy" grades). The FSRS scheduler uses these two signals to set intervals.
Interface languages are limited
Origa's interface is currently available in English and Russian. Additional languages are planned but not yet shipped.