How do I make a scanned PDF searchable?
A scanned PDF is a stack of photos. You can read it, but your computer can’t: search finds nothing, and selecting text grabs the whole page as a picture. OCR fixes that by recognizing the letters on each page and storing them as text the computer understands.
Choose your file, tick the languages printed in it and keep Searchable PDF as the output. The tool renders each page at high resolution, recognizes the text and then places an invisible text layer exactly over the words on the original page. The result opens in any PDF reader. It looks identical to your scan, but search, text selection and copy now work, and file search on your computer can find it by its contents.
Pages that already contain real text, for example a typed cover page in an otherwise scanned file, are left as they are, so you never get the same words twice.
Tip: Tick only the languages that are really in the document. Each extra language makes recognition a little slower and gives the engine more chances to guess the wrong word.
How do I convert a scanned PDF to text?
Pick Text file (.txt) as the output. You get one plain text file with the words from every page, separated by a blank line between pages. That’s handy when you need to quote a scanned letter, translate a document, or move figures from an old report into a spreadsheet.
If you want a document you can edit with its layout, the next step is converting the searchable PDF to Word. For files that were never scanned, such as a PDF exported from Word, OCR isn’t needed at all: extracting the text directly is faster and exact.
How accurate is OCR on scans and phone photos?
On a clean, straight scan of printed text, recognition is usually very good. Accuracy drops when the input is hard to read, so it helps to know what the engine struggles with:
- Low resolution. Scan at 300 DPI. Text that looks blurry when you zoom in will be guessed.
- Tilted or curved pages. Photos of a book spine or a crooked phone shot confuse line detection. Straighten the page first.
- Handwriting. The engine is trained on printed type. Neat block letters may work, joined handwriting mostly won’t.
- Stamps, watermarks and colored backgrounds. Text on top of them is harder to separate from the background.
- Tables and multiple columns. The words are found, but the reading order in a .txt file may jump between columns.
Always check names, numbers and amounts in the result before you rely on them.
Is it safe to OCR confidential documents online?
It is here, because the document is never uploaded. Most online OCR services send your file to a server, run the recognition there and keep a copy for some time. This page downloads the recognition engine and its language data to your browser instead, and the whole job runs on your own phone or computer.
You can verify it: open your browser’s developer tools on the Network tab while you run OCR. You’ll see the engine files arrive the first time, but your PDF is never sent anywhere. That matters for the papers people usually scan: ID cards, tax returns, bank statements, medical letters and signed contracts.
Which languages can I recognize?
English and Indonesian are available, and you can tick both for documents that mix them, such as an Indonesian contract with English clauses or an English invoice addressed to a company in Jakarta. The models are the accurate “best” versions of Tesseract’s language data, packed in a smaller integer format, so they download quickly and stay precise.