How will OCR read your document?
Yomikurabe runs a single document through several OCR (text recognition) engines at once, then lines up the recognized text and processing times so you can compare them. Find out which engine suits your documents, right on your phone.
Yomikurabe is designed with Japanese documents in mind, including handwriting and vertical text.
■ Just snap or pick
- Take a photo, choose one from your gallery, or pick a page from a PDF
- Every engine receives the exact same image, so the comparison is fair
■ On-device engines, no sign-up needed
No account or API key required. Your images never leave your device.
- ML Kit Japanese
- RapidOCR (small / medium)
- NDLOCR-Lite
- PP-OCRv5 server
■ Try cloud OCR too (optional)
- Google Cloud Vision
- Gemini
Enter an API key you obtained yourself on the settings screen to use them. Images are sent to Google services only when you select a cloud engine. Charges are billed under your own Google Cloud / Gemini account; the app shows an estimated cost based on a price table.
■ Line them up and compare
- Switch between engines to compare the recognized text
- See recognized regions outlined on the image, with zoom in and out
- View the processing time for each engine
■ Keep and export your results
- Save runs to history (stored on your device)
- Export measurements as CSV and recognized text as a text file
■ Who it's for
- Anyone who wants to see how OCR engines differ before using OCR at work
- Anyone curious how their forms, handwriting, or vertical text will be read
- Anyone who wants to compare on-device OCR with cloud OCR
■ Please note
- You can select up to 3 engines at a time
- Results vary by engine and image. Always check the results yourself for important documents
- The app is free and shows ads
- Requires Android 12 or later
Run one document through several OCR engines and compare results side by side