How does OCR text extraction work on MakeOverPDF?

Read the text out of a scanned document. MakeOverPDF runs Tesseract on the server and returns: A .txt file of the recognised words, downloaded immediately.

Is OCR text extraction free and private to use?

Yes. MakeOverPDF requires no user account, watermarks nothing unless you use the watermark tool, and deletes your uploaded document from the private working directory as soon as the request finishes.

What are the file size and page limits for OCR text extraction?

Each page is rendered and recognised in turn, and the words come back as characters. The shapes, stamps and pictures around them are not described — this reads words, it does not narrate the page. Recognition follows the quality of the scan: small print, poor contrast or a tilted photograph produces wrong words, and the result never says which ones it is unsure about. The recogniser knows the languages the administrator has installed, and no others; text in a language with no data on this server will not come back correctly. When the recogniser finds nothing at all, the tool says so instead of handing back an empty file — a document whose pages are blank, unreadable or in an uninstalled language fails with that explanation. A long document is refused up front, naming its real page count: there is a page ceiling per run, and the remedy is to split the document and recognise the parts. A document that already has a text layer is better served by PDF to text, which reads the existing characters exactly instead of recognising them again from pixels.

Available now Reliable

OCR text extraction

Read the text out of a scanned document.

Run this tool

The file is processed on this server and deleted as soon as the response is sent. Nothing is stored, and the result is returned directly to you rather than parked behind a link.

The document

Choose a scanned PDF to read Up to 25 MB. Each page is rendered and recognised, one page at a time.

Recognition

Language

The language the installed Tesseract models are set up for. The result is only as good as the model that was installed for this language.

Eng (the only language this installation has)

Resume from page

Leave this blank to recognise the whole document. When a previous run only finished partway, enter the last page it completed so this run starts after it instead of redoing the earlier pages.

This reads what the recogniser sees in the pixels. For a document that already has a text layer, PDF to text is the faster and more accurate tool.

How to use OCR text extraction

Select your document: Choose the file you want to process with OCR text extraction. Your upload is validated immediately against strict format and size limits.
Configure ocr text extraction options: Adjust any page ranges, compression targets, passwords, or layout settings required for your task. Read the text out of a scanned document.
Run and receive your result: Submit the form to run Tesseract directly on this server. Output: A .txt file of the recognised words, downloaded immediately. Temporary input files are deleted automatically when processing completes.

Frequently asked questions

How does OCR text extraction work on MakeOverPDF?: Read the text out of a scanned document. MakeOverPDF runs Tesseract on the server and returns: A .txt file of the recognised words, downloaded immediately.
Is OCR text extraction free and private to use?: Yes. MakeOverPDF requires no user account, watermarks nothing unless you use the watermark tool, and deletes your uploaded document from the private working directory as soon as the request finishes.
What are the file size and page limits for OCR text extraction?: Each page is rendered and recognised in turn, and the words come back as characters. The shapes, stamps and pictures around them are not described — this reads words, it does not narrate the page. Recognition follows the quality of the scan: small print, poor contrast or a tilted photograph produces wrong words, and the result never says which ones it is unsure about. The recogniser knows the languages the administrator has installed, and no others; text in a language with no data on this server will not come back correctly. When the recogniser finds nothing at all, the tool says so instead of handing back an empty file — a document whose pages are blank, unreadable or in an uninstalled language fails with that explanation. A long document is refused up front, naming its real page count: there is a page ceiling per run, and the remedy is to split the document and recognise the parts. A document that already has a text layer is better served by PDF to text, which reads the existing characters exactly instead of recognising them again from pixels.
How reliable is OCR text extraction?: Reliable (Tesseract): Requires the optional Tesseract engine. Text quality follows scan quality.

How it works

Engine: Tesseract
Category: PDF Utilities
Reliability: Reliable
Output: A .txt file of the recognised words, downloaded immediately

What to expect

Requires the optional Tesseract engine. Text quality follows scan quality.

Each page is rendered and recognised in turn, and the words come back as characters. The shapes, stamps and pictures around them are not described — this reads words, it does not narrate the page.
Recognition follows the quality of the scan: small print, poor contrast or a tilted photograph produces wrong words, and the result never says which ones it is unsure about.
The recogniser knows the languages the administrator has installed, and no others; text in a language with no data on this server will not come back correctly.
When the recogniser finds nothing at all, the tool says so instead of handing back an empty file — a document whose pages are blank, unreadable or in an uninstalled language fails with that explanation.
A long document is refused up front, naming its real page count: there is a page ceiling per run, and the remedy is to split the document and recognise the parts.
A document that already has a text layer is better served by PDF to text, which reads the existing characters exactly instead of recognising them again from pixels.

Privacy

Your document is written to a private directory outside the web root, processed by a local executable, and deleted when the request ends. There is no upload URL, no account, and no third-party service involved.

OCR text extraction

Run this tool

How to use OCR text extraction

Frequently asked questions

Related PDF tools