How to OCR a Scanned PDF Online for Free

Direct answer: You can OCR a scanned PDF online for free with the OCR PDF tool: it reads every page of a scanned PDF and rebuilds it as a searchable PDF, keeping each page image unchanged while adding an invisible text layer aligned over it. The result looks exactly like the original scan, but its text is now searchable, selectable, and copyable, as if the document had been born digital.

That transformation changes how a scanned archive can be used. Before OCR, a scanned contract, book, receipt, or report is a set of pictures: you can read it by eye, but you cannot search for a name, copy a paragraph, or find a clause without turning every page. After OCR, the same document behaves like a normal PDF for searching and copying, while still showing the original page images, including handwriting in the margins, stamps, signatures, and the original layout.

What OCR does to a scanned page

Optical character recognition reads the shapes on a page image and converts them into characters, words, and lines of text. The tool then places that recognised text in an invisible layer that sits exactly over the page image, aligned with the printed words beneath it. When you select text in the finished PDF, you are selecting the invisible layer, and when you search, you are searching that layer, even though what you see on screen remains the original scan.

Keeping the page image unchanged is a deliberate choice with real benefits. Signatures, stamps, handwritten notes, diagrams, and the original typography all remain visible exactly as they were scanned, so the document still looks like the original evidence. At the same time, the text layer gives you the digital conveniences of a born-digital file. If the recognition makes a mistake on a particular word, the original image is still there to check against, which is far better than a conversion that replaces the image with plain text and loses the evidence of what was actually on the page.

Languages, scan quality, and what OCR cannot do

Before running OCR, pick the language of the document: English, Spanish, or Urdu. Language choice matters because letter shapes, word patterns, and script rules differ, and the correct model will read far more accurately than a wrong one. For Urdu, printed Naskh text reads best, while Nastaliq is best-effort, because its connected, sloping forms are harder for recognition to follow. If a document mixes languages, run it with the dominant language first and check the minority-language pages closely afterwards.

Scan quality is the single biggest factor in accuracy. Clean scans at 200 DPI or above, with straight pages, good contrast, and sharp printed text, read near-perfectly in most cases. Faint pages, skewed pages, pages with heavy show-through from the reverse side, and very low-resolution images will have more errors, because the recognition engine has less clear information to work from. Handwriting is not supported: handwritten pages will produce errors or nonsense, and should be transcribed by hand rather than passed through OCR with the expectation of a clean result. Knowing these limits before you start will save you from trusting a text layer that looks complete but contains silent mistakes on the hardest pages.

Step by step: running OCR on your scanned PDF

Start by confirming that your file actually needs OCR. A born-digital PDF already has a text layer, and running OCR on it will not improve it; you can test this by trying to select text in the original file. If text selects cleanly, the file is already searchable. If the pages behave like pictures and no text selects, the file is a scan and OCR is the right step. That quick test prevents wasted time on documents that were never scans in the first place.

Load the scanned PDF, choose the language, and start the run. Progress is shown page by page, and each page's text appears as it finishes, so you can watch the document build up rather than waiting blindly for a single finished file. A short document may finish quickly, while a long one takes longer: a 50-page scan can take minutes, and you should keep the tab open until it completes, because closing the tab will interrupt the work. Plan that time into your task, and do not start a long OCR run moments before you need to close your computer. If a run is interrupted, restart it rather than trying to use a partial output, because a partial text layer can look complete while missing whole pages at the end, and a fresh full run is the only way to be certain every page was processed.

• Check whether text already selects in the PDF; if it does, OCR is not needed.
• Open the free tool and load your scanned PDF.
• Pick the language before running: English, Spanish, or Urdu.
• Start OCR and watch progress page by page as text appears.
• Keep the tab open until the full document finishes, especially for long scans.
• Download the searchable PDF and test search, selection, and copying.
• Proofread important pages against the original page images.

Proofreading the text layer after OCR

OCR output should be proofread, especially where accuracy matters. Numbers are the highest-risk area: a misread digit in a date, an amount, or a reference number can change the meaning of a document completely, while a misread ordinary word is usually obvious from context. Names, technical terms, and words in a second language are also worth checking, because recognition may substitute a familiar word for an unfamiliar one. Read the text layer alongside the page image for any passage you intend to quote, copy into another document, or rely on for a decision.

Proofreading is also where you decide whether a rescan would help. If one page is consistently poor while the rest are clean, rescanning just that page at higher resolution and better contrast, then re-running OCR, will often fix the problem far more reliably than manually correcting dozens of errors. If the whole document is faint or skewed, a better scan of the whole document is worth the effort, because a clean scan is the foundation that every later step depends on. Keep the original scan even after a successful OCR run, as the authoritative copy, and treat the searchable PDF as a working copy for searching and copying. For a document you will quote often, keep a short proofreading note listing the pages you checked most closely and any words you corrected, so that a later reader knows which passages were verified against the image and which were accepted as recognised, and so that a future re-run can focus on the pages that were hardest the first time.

What to do with a searchable PDF afterwards

Once the text layer exists, the document opens up to ordinary PDF workflows. You can search for a clause across a hundred pages in seconds, copy a paragraph into an email or a report, and use text-based tools that were useless on a picture-only scan. For example, a long scanned report becomes a document you can navigate by search terms rather than by flipping pages, and a scanned archive becomes a collection you can actually query, rather than a set of images you must remember by sight.

That new capability also connects to other PDF tasks. If you need the plain text itself, rather than a searchable PDF, you can extract text from the OCR output. If you need data from tables inside the scan, remember that table extraction needs a text layer to work from, so OCR is the necessary first step before table tools can see the content at all. You can read the plain-text workflow in our guide to getting text out of a PDF, and explore further PDF guides on the blog archive.

For archives, give the searchable PDF a clear filename and keep it alongside the original scan, rather than replacing the original. Future readers may want to verify a passage against the original image, and keeping both files preserves that option. A short note in the filename, such as searchable, will tell future readers which copy to search in and which copy to treat as the original evidence.

Frequently Asked Questions