Free PDF OCR Tool

What is PDF OCR?

PDF OCR, optical character recognition, reads the text in a scanned PDF's pictures and adds it to the PDF as real text.

A scan is only a picture of a page. You can't search it, copy from it or read it aloud until its text is read.

What you get

Each PDF comes back 2 ways:

• A searchable PDF – the same pages, each read page with its text laid under it. Search it, select it and copy it in any PDF reader.
• The text – every page's text, in reading order, to copy or download as a TXT file.

Pages that already have text, like a PDF saved from Word, are kept as they are. So a PDF with a few scanned pages only gets those read.

How it reads a bad scan

Before it reads a page, the tool looks at it and cleans it up only where that helps:

• A small scan is enlarged – to 3,000 pixels on its long side. On the forms, that alone took the score from 54% to 75%.
• Uneven light is evened out – a shadow, a yellowed page or a dark corner is divided out by the page's own background.
• A dark page is flipped – light text on dark, like a dark mode screenshot, becomes dark on light.

A sharp, evenly lit page is read as it is. On 4 such pages, 300 dots an inch, it scored 94%.

Compared with CloudConvert

I ran 10 real scanned business forms from FUNSD, a public research set, through this tool and through CloudConvert's OCR, on September 30, 2026.

Then I ran the same forms again, made to look like phone photos: a shadow, noise and heavy JPEG.

The score is the share of each page's words read right, counting missed and extra words:

The forms were scanned at a low resolution, 88 dots an inch, like many office scans. CloudConvert read them as they were.

This tool enlarges a small scan first and evens out the light, which is where the difference comes from.

19 languages

Pick the language your text is in: English, Spanish, Portuguese, French, German, Italian, Dutch, Swedish, Polish, Turkish, Russian, Arabic, Hindi, Chinese, simplified, Chinese, traditional, Japanese, Korean, Vietnamese or Indonesian.

Each one is downloaded once, 1 to 3 MB, the first time you pick it. Your browser keeps it after that.

Nothing is uploaded

Most online OCR tools send your file to their server. This one reads it in your browser, so contracts, IDs and medical letters never leave your computer.

The first time, the reader downloads, about 4 MB, plus your language. After that it starts at once.

What it can't read

Some pages don't come out well:

• Handwriting – it reads printed text. Neat block capitals sometimes work.
• Tables – each cell's text is read, but the TXT file doesn't keep the columns.
• Tiny or blurry text – very small print or a shaky photo loses letters.
• A password – a locked PDF has to be saved without it first.

The confidence under each file says how sure the reader was. Below 60%, check the text.

Who is the PDF OCR tool for?

It's for anyone with scans to search:

• PDF to Markdown – people turning documents into notes for AI
• Compress PDF – anyone emailing big scans
• Merge PDF – people putting scanned pages together
• Speech to text – researchers turning interviews into text too

Freelancers, lawyers and small offices use it for old contracts and invoices.

More free PDF and text tools

Here are my other tools for scans and documents:

• Image to text – the text in any screenshot or photo
• PDF to Markdown – a PDF that already has text, as clean Markdown
• Compress PDF – a smaller PDF, same pages
• Merge PDF – several PDFs as one
• PDF to JPG – each page as a picture

They're all free, in your browser, with no watermark.