Skip
  • TOP　>　
  • News　>　
  • Local OCR Added to TEI Tools

2026.10.01

  • Research

Local OCR Added to TEI Tools

We have added “Local OCR,” a free text recognition application, to TEI Tools (https://tei-tools.toyobunko-lab.jp/), the Toyo Bunko’s browser-based suite for creating, editing, and publishing TEI/XML texts.

Local OCR is a free desktop application for macOS and Windows. Just install it and you’re ready to go, with no programming or technical setup required. Open an image of your source, click a button, and the text is recognized.

Choose the right OCR engine for each language and script
Local OCR comes with several recognition engines, each with its own strengths. You can switch between them in the interface to suit your material:

  • Modern printed Japanese: NDLOCR Lite
  • Premodern Japanese, including cursive script (kuzushiji) and kanbun: NDL Kotenseki OCR Lite
  • Classical and modern Chinese, and multilingual material: PaddleOCR-VL
  • Tibetan: Yigdzin-1

You can also run two engines on the same page and compare their output side by side.

Your images stay on your computer
All recognition runs locally, so no image or recognized text is sent to an external server. This makes it safe to use with unpublished materials and works still under study. Images from digital archives can also be loaded directly from a IIIF manifest URL.

Straight to TEI/XML
Results can be saved as plain text (.txt) or as TEI/XML (.xml) with facsimile zones that link the text to its position in the image. Together with the editing and publishing tools in TEI Tools, this lets you go from page image to transcription, critical editing, and online publication.

Local OCR is developed by Satoru Nakamura (The University of Tokyo) and released as open-source software under the MIT License.

Back