OCR PDF

OCR works by comparing the shapes on a page with a language model. Choose the main language before you tap Run OCR. The picker includes English, Arabic, Spanish, French, German, Italian, Portuguese, Dutch, Russian, Hindi, Chinese, Japanese and Korean.

Use one language when the document is consistent. For a page with mixed content, you can select more than one language in the picker. Keep the selection focused when you can, because a clear match gives the reader fewer possible character shapes to compare. As an example, an English contract with a short French clause reads best when English leads and French is added, rather than loading every script at once.

Language choice cannot repair a blurred scan or restore ink that is missing. It helps the reader interpret the marks that are already visible. If the first result looks wrong, select the correct model, use the clearest source page and run it again.

Choose OCR PDF when you want the page image to stay visible. The tool renders each page, recognizes the words and creates a new PDF with searchable text. The finished file uses the original page as the visual reference, so you can search or highlight words without turning the document into a plain text page.

Choose Plain text when the layout does not matter. The recognized words appear in the editor and are also saved as a text file, which is useful for notes, translation, a spreadsheet or another writing tool. Plain text will not keep fonts, images, columns or table borders.

These outputs serve different jobs. OCR PDF is better for archiving and sharing a document that should still look like the scan. Plain text is better when you need to reuse the content and plan to format it yourself.

A searchable PDF keeps the scanned picture of each page and adds a text layer beneath it. That layer is what lets you find a word, copy a line or let another program read the file. Making a PDF searchable does not change how the page looks; it changes what the page holds underneath the image.

To check the result, open the finished file and try to select a line or search for a word you can already see on the page. If the words highlight, the text layer is in place and the scan is now a searchable PDF. If nothing selects, the page is still only a picture, so run the pass again with the matching language and the sharpest copy you have.

A clean scan with strong contrast gives the reader the best starting point. Straight pages, dark type and even lighting are easier to recognize than a tilted phone photo, faded carbon copy or page with deep shadows near the binding.

Resolution counts as much as focus. A scan around 300 dots per inch gives the reader enough detail on each letter, while a small or heavily compressed image leaves the edges too rough to match. If a page came from a phone, hold the camera square to the paper and fill the frame, because an angled shot stretches the letters and lowers the result. Straightening a crooked page and cropping away the desk or wide margins around the document both give cleaner input for the same effort.

Handwriting, decorative fonts, small type and crowded tables can produce mistakes. OCR reads the visible shapes, so it may confuse characters that look alike or lose the order of words in a complex layout. Always compare important names, dates, totals and reference numbers with the source image.

If the result is poor, try the sharpest source you have, choose the matching language and run the page again. OCR cannot recover detail that the scan never captured.

OCR is for PDFs where the words live inside a picture: scans from a copier, faxes and photos of paper. If you try to select a word and nothing highlights, the file probably needs OCR. A PDF exported from a writing app usually already has a text layer, so PDF to text can pull its words out directly without a recognition pass.

A file can contain both kinds of page. One page may have selectable text while another is only an image. OCR is useful when you need a searchable result for the scanned pages, but the final quality still depends on what each page contains.

Photographing a page with a phone is the most common way people scan today, and those shots almost always need OCR before the words can be selected. Once the recognition pass finishes, a photographed page behaves like any other scan to text result: you can search it, copy from it or send it on to a translation tool.

Add a group of scans and choose the language and output once. Each PDF is read in turn with the same settings. A result is offered for each source file, and the editor can bundle the completed files into one ZIP so you do not need to download every item separately.

Batch work uses your device memory and processor. A long scan or a large group may take longer than one short page. If one file cannot be read, the rest of the batch can still finish, so you do not have to restart every document because of one bad input.

Keep the source filenames clear before you start. The completed PDF or text result keeps the base name and adds the output type, which makes a group easier to match back to the originals for later review today. Stay on the page while the progress bar is moving, especially on a phone, and wait for the completion message before closing the tab.

If a PDF opens without a password, it can go straight to the language and output controls. If it needs a password just to open, the editor shows an unlock box beside the file. Enter the password there, and the file is unlocked in your browser before OCR starts. A PDF that only limits printing or copying may open without a prompt because it is not locked against reading.

Scanned contracts, identity documents, medical letters and financial records can contain details you would not want stored on an outside service. This tool reads the pages in your browser, so the source PDF is not uploaded for recognition. The password, when one is required, is entered in the same local flow.

Once the searchable file or text is downloaded, treat it like the original. OCR changes what you can search and copy, but it does not make a private document safe to publish. Check the words and the visible page for names, addresses, account numbers and other details before sharing.

Use Compress PDF if the searchable result is too large to send. Join it with another document using Merge PDF, remove pages with Remove pages, or reorder the result with Organize PDF. If you only need the recognized words, use Plain text instead of making a second conversion.

When you pick Plain text, the words drop straight into a note, an email or a spreadsheet cell without a second export. Scanned tables rarely keep their rows, so paste the text and line the columns back up by hand. Proofread names, dates and totals every time, because a single misread digit can change what a scanned record means.

Yes. Open the page in a current mobile browser, add the scan, choose its language and tap Run OCR. The reader uses the phone while it works, so a multi-page file may take longer than it would on a computer. Keep the tab open until the download is ready.

• Making a scanned contract or lease searchable so you can find a clause.
• Copying text from a receipt, invoice or form into another document.
• Turning a photographed page of a book or article into editable text.
• Archiving paper records as searchable PDFs.
• Checking a scanned report before sending it to a text or translation tool.