Scribe OCR is a free, open-source web application for recognizing text in scanned documents, proofreading OCR output, and producing fully digitized documents. The project lives on GitHub with about 810 stars under the AGPL-3.0 license, was updated in 2026, and runs as a live service at scribeocr.com.
Most OCR tools either add a roughly positioned invisible text layer over a scan or make correcting recognition errors painful. Scribe OCR focuses on accuracy and correction: recognized text is positioned precisely over the input image, so proofreading is significantly faster than in typical tools, and existing OCR data — including Tesseract HOCR files — can be imported and fixed. It can add an accurate searchable text layer to PDFs as an alternative to tools like Adobe Acrobat, or produce text-native, ebook-style PDFs that genuinely replicate the original document.
One detail worth knowing is that the GitHub repository contains the code for the user interface; the actual character recognition is performed by the well-known Tesseract OCR engine. What Scribe OCR adds on top is the workflow around recognition — an editing environment where recognized words stay aligned with the scanned image, making error correction fast, and export paths that produce genuinely reflowable digital documents instead of image-plus-hidden-text hybrids. That focus on corrected output is what distinguishes it from quick online OCR converters.
Scribe OCR is free. The web application at scribeocr.com costs nothing to use, and the source code is open under the AGPL-3.0 license for self-hosting.
Scribe OCR is for archivists, librarians, researchers, and anyone digitizing books or document collections who cares about corrected, truly digital text rather than a quick-and-dirty searchable scan.