OCR 2.0 · local document workflow

Recognize images and scanned PDFs, then keep the text editable.

Select pages, correct orientation, and enhance low-contrast scans. The OCR engine, files, and results stay on this device.

  1. 1 · Choose document
  2. 2 · Recognize
  3. 3 · Review pages
  4. 4 · Export

Choose or drop images / PDFs

    The self-hosted OCR engine and selected language model load on first use. Models can be cached for offline use; files are never uploaded.

    Recognized text will appear here and can be edited before export.

    PDF with the scan layout

    Keeps the processed scan and original OCR text layer, without manual corrections. Use Print / PDF transcript above for corrected text.

    PDF to WordMerge & split PDFMy tool workspace

    How to improve local OCR accuracy

    Start with the correct orientation. Try Grayscale + high contrast for faded scans, or Black and white threshold for clean dark print on a plain background. Thresholding is not suitable for photos, colored tables, or light text.

    The page selection applies to every selected PDF and each run is limited to 10 pages. Page confidence helps locate likely review work; it does not prove the text is correct. Compare legal, medical, financial, or identity documents line by line with the source.

    Searchable PDF export keeps a page image with a recognized text layer. Complex tables, handwriting, curved pages, and mixed languages can still require specialist OCR or manual correction.