Make a scanned contract searchable
Turn a whole scanned contract into a searchable PDF so clauses can be found with Ctrl+F and copied later, while it still looks like the original.
Loading…
Guide
PDF OCR handles scans and image-only PDFs where the text cannot be selected: every page is first rendered to a 300 dpi image, recognised by the tesseract engine inside the browser, and then exported in one of three shapes — a “searchable PDF” that keeps the original page picture and adds an invisible text layer, a TXT file whose pages are separated by — Page n — , or a DOCX where each page is a group of paragraphs separated by a page break.
Updated 2026-09-094 min read
PDF OCR handles scans and image-only PDFs where the text cannot be selected: every page is first rendered to a 300 dpi image, recognised by the tesseract engine inside the browser, and then exported in one of three shapes — a “searchable PDF” that keeps the original page picture and adds an invisible text layer, a TXT file whose pages are separated by — Page n —, or a DOCX where each page is a group of paragraphs separated by a page break.
A PDF that already has a text layer (selectable, and findable with Ctrl+F) does not need OCR: “PDF to Word” or “PDF Extract Text” read the text layer directly, which is faster and more accurate. When the input is a set of standalone images, use “Image OCR”.
<name>_searchable.pdf, <name>_OCR.txt and <name>_OCR.docx. The first searchable PDF also loads the bundled Noto Sans SC CJK font (about 10 MB, once, then cached).| Input | Output | Notes |
|---|---|---|
scanned-contract-12p.pdf |
scanned-contract-12p_searchable.pdf, reporting “2,340 words embedded” |
Looks identical to the original but now has selectable text |
| A TXT download of the same file | “ — Page 1 — ” … plain text in sections |
Edits made in the text box are written into the TXT |
| A text-based (non-scanned) PDF | It can be recognised too, but far more slowly than direct extraction | “PDF to Word” is the better choice |
Turn a whole scanned contract into a searchable PDF so clauses can be found with Ctrl+F and copied later, while it still looks like the original.
Recognise scanned old reports and papers into TXT or DOCX and edit them again in a word processor.
Combine several photographed pages into one PDF with “Scan Enhance”, then recognise it here into a searchable version.
The appearance is exactly the same — the original scanned picture. The difference is the invisible layer of text on the page, which lets a reader select, copy and search, and lets a screen reader read it aloud.
The text layer has to embed CJK glyphs, so the tool uses a bundled Noto Sans SC font (about 10 MB) that is downloaded from this site once and then cached, after which it reports “the CJK font is cached locally”. The PDF it produces embeds only the subset of glyphs that were used.
No. After “Cancel recognition” the pages already recognised are kept and can be downloaded straight away as TXT, DOCX or a searchable PDF, covering only the pages that finished.
Rendering and recognition both happen locally in the browser, and the page content is never uploaded. The recognition engine and the Simplified Chinese/English language data are self-hosted by this site and downloaded on first use; the Traditional Chinese and Japanese language data come from the tesseract.js default public CDN; generating a searchable PDF downloads the Noto Sans SC font (about 10 MB) from this site once and caches it. Those requests fetch only the engine, language data and font files — never your document.
Updated 2026-09-09
OCR scanned PDFs page by page; output a searchable PDF (invisible text layer), TXT or DOCX
For scans and image-only PDFs: each page is rendered at 300dpi and then recognized, and you can output a searchable PDF / TXT / DOCX