Extract Text From Scanned PDF Online Free: Real Options

You can extract text from a scanned PDF online for free in three ways: open it as a Google Doc in Google Drive, use a paid converter's free tier, or use a site that asks for no account. Each returns something different and stops at a different limit. Free software on your own computer avoids the upload.

This post is a survey, written so you can pick a route in a few minutes. For each one it says what lands in your downloads folder, which number stops you, and whether you need an account. It quotes each product's own documentation and makes no claim about which one reads a page best. Only your own pages can answer that.

What does "free" hand you when you extract text from a scanned PDF online?

Free routes hand you one of four things: a word-processing document, a searchable PDF, a plain text file, or a choice of formats. Which one you get matters more than the word "free", because a tool can read your scan perfectly and still return a file that does not suit the job.

Start from the job, then look for the route whose output matches it.

• You want to paste two paragraphs into an email or a brief. Any output works, as long as you can select the words. A plain text file brings the least baggage with it.
• You want to search the document and keep it looking like the original. You need a searchable PDF, which is the scanned page with recognized text behind it.
• You want to rewrite the document. You need a word-processing file.
• You want the figures from a table. You need an export that keeps rows and columns apart, such as a spreadsheet.

The second thing "free" decides is the limit. Every free route has one, and they are not the same kind. One route talks about megabytes, another about pages per document, another about documents per day, and local software asks for your time in place of a quota. Count your pages and look at the file size before you choose, since those two numbers rule out half the options at once.

If you are not yet sure the file is a scan, the one-minute test in why you can't copy text from a PDF settles it. A PDF that already has text behind its pages needs no OCR, only copy and paste.

Opening the scan as a Google Doc from a Drive account you already use

Google Drive turns a scanned PDF into an editable Google Doc. You upload nothing new if the file already sits in Drive, and there is no separate product to sign up for. What comes back is a document of recognized words, not a PDF, so the picture of the page stays behind in the original file.

Google's help page on converting PDF and photo files to text gives the whole procedure in one line: "Right click a file. Click Open with and then Google Docs." The same page says you can convert "PDFs (multipage documents) or photo files (.jpeg, .png and .gif)".

The free limit here is file size, stated as a guideline. Google's page says "The file should be 2 MB or smaller." It also asks that "Text should be at least 10 pixels high" and that "Documents must be right-side up." A scan made in color at a high resolution passes 2 MB within a few pages, so check the size of yours first.

The second limit is structure. In Google's words, "Lists, tables, columns, footnotes, and endnotes are not likely to be detected." A plain letter is inside that guidance. A fee schedule or a footnoted opinion is not.

This route suits a short, clean, small file when all you want is the wording. The post on when a Google Drive OCR alternative suits client files goes through Google's published limits and its terms in detail, so this one stops here.

The free plan of a paid converter, with two examples you can check

A paid converter's free plan lets you run a small amount through the same product its customers pay for. You usually sign in, and the cap is on pages, documents or what you may do with the result. Read the company's own page for the cap, because each one draws the line somewhere different.

Example one: Schema Law's Free plan

The Free plan on Schema Law's pricing page is one document a day, with up to three pages per document. You sign in with a Google account and upload a PDF, PNG or JPEG of up to 20 MB in the browser. The pages are read into text and tables, and you export them.

The export is where this example differs from a single-output tool. Every plan, Free included, exports all eight of the formats listed on the homepage: Word, Excel, CSV, JSON, searchable PDF, plain text, Markdown and HTML. So a three-page order can come back as a text file for quoting and as a searchable PDF for the file, from one upload.

The limit bites on length. A two-page notice fits. A seven-page agreement is over the three-page allowance for one document, and one document a day means you cannot work through a stack this afternoon. Treat the Free plan as a way to extract text from a short scan, or to try the product on a real page.

Example two: Adobe's online OCR tool

Adobe publishes a browser page for its Acrobat online OCR tool. The steps on that page are to select or drop in a PDF, wait while the tool applies OCR, and then, quoting the last step, "Sign in to download your newly searchable PDF or get a link to share it."

What you get back, then, is a searchable PDF. You can select and copy words from it, but it is not a text or Word file. The page's own answer on editing reads: "With a free 7-day trial of Acrobat Pro, you can edit recognized text, reuse content, and preserve formatting and layout in your PDF after OCR is applied." So the limit on this route is what you may do with the result, more than a page count.

These two examples show why a general statement about free tiers only gets you so far. The post on what free tiers limit when you convert a scanned PDF to Word lists the kinds of cap and how to test one before you depend on it.

What can you tell about a no-account converter site before you upload?

A no-account converter is a web page with an upload box and a download button. Before you upload you can usually tell three things from the page itself: what file type it promises back, what size or page cap it states, and whether a named company stands behind it. Anything else you learn only by trying.

This post describes these sites as a category and names none of them, because they come and go and their terms change. As a category they share a few traits.

• What you get back. Usually one format, chosen before you upload. Many offer plain text, which is the most direct answer to "extract text". Some return the words in a box on the page for you to copy.
• Where the limit bites. The cap is often stated in megabytes or pages per file, and sometimes in conversions per hour. Some sites state no cap and stop partway through a long file. Count the pages you got back.
• No account. That is the attraction, and it has a cost. With no account there is no list of what you uploaded and no button to remove it.

For a published opinion or a blank form, none of that matters much. For anything a client gave you, it matters a great deal, and the question is bigger than this survey. The post on whether free online OCR is safe for legal documents explains how such sites earn money and gives a short check to run on any of them.

One practical tip applies whatever the document. Run a page you do not care about first, one page longer than the stated cap if there is one. You learn in a minute whether the site refuses the file, cuts it short, or handles it.

Free OCR software on your own computer as the no-upload comparison

Software installed on your own computer is not an online route, but it belongs in the comparison because it is free and the scan never has to be uploaded to a website. The price is setup. The best-known free tools are run by typing commands, which suits some offices and not others.

Tesseract is the usual starting point. The Tesseract documentation describes it as "an open source text recognition (OCR) Engine, available under the Apache 2.0 license", and adds that "Tesseract doesn't have a built-in GUI". Its command-line guide shows the basic form, tesseract imagename outputbase, and says "The default output format is text."

There is a catch for anyone holding a scanned PDF. Tesseract's page on input formats says plainly: "Tesseract does not support reading PDF files." It reads images such as PNG, JPEG and TIFF, so a PDF has to become page images first, or go through a second tool.

OCRmyPDF is one such tool. Its introduction calls it "a Python application and library" that makes scanned image PDFs searchable, and says it "uses Tesseract, a widely available open source OCR engine, to perform OCR." Its cookbook gives a command that writes a companion text file beside the new PDF:

ocrmypdf --sidecar output.txt input.pdf output.pdf

That one line returns both a searchable PDF and a text file. The introduction is also frank about limits, saying the tool "is incapable of recognizing handwriting".

Neither project sets a page quota in the passages quoted here. What stops you is the installation and the command line. If someone in the office is comfortable with both, this is the free route with no upload step. The post on what makes a converter private compares local and hosted tools more fully.

The four free text-extraction routes in one table

The table puts the four routes next to each other on the points that decide a choice: what you get back, where the free limit sits, whether you need an account, and what to look up before a client file goes in. Each product row repeats only what that product's own page says.

Two cautions about reading the table. It compares what comes back and what stops you. It does not rank how well each route reads a page, and no vendor's description can tell you that. And every cell about another company's product can change, so open the linked page on the day you rely on it.

For the Schema Law row, the company's own statements for every plan include "Your documents are never used to train an AI model", "Documents stay inside your own account" and "Delete any document, at any time". Those are the company's words, and you should weigh them as you would any vendor's.

Which free route to use for which scanned PDF

Match the route to three facts about the file: how long it is, what you need back, and whose document it is. A short public document with plain paragraphs can go almost anywhere. A long file, a table, or a client's record each narrows the field quickly.

A few follow-up questions come up often.

Can you split a long scan to fit a free cap? You can, and people do. Every split is a chance to lose or reorder a page, so number the pieces by page range and count the pages again when you join the results.

Is it worth trying two routes on the same page? Yes, when the document is not confidential. Ten minutes with the same hard page in two tools tells you more than any description. The post on how to test OCR software sets out a fair way to run that comparison.

What if nothing may be installed and nothing may be uploaded? Then free and online are both off the table for that file, and the question belongs to whoever sets the firm's rules on tools. The post on OCR with no install covers what a browser-only tool involves.

Three checks on free extracted text before you rely on it

Text from any free route needs three checks before it goes into your work: that every page came back, that the reading order is right, and that names and numbers match the scan. They take a few minutes and apply equally to a Google Doc, a searchable PDF and a text file.

1. Count the pages. Compare the page count of the scan with what came back. A cap that cuts a file short without warning is the worst failure of a free tool, because the result looks finished.
2. Read the last page and one from the middle. A sidebar, a two-column layout or a caption box can be read out of order. Follow a few lines against the scan to see that sentences run on as they do on paper.
3. Check every name, date, amount and section number you plan to use. A wrong digit in a figure reads as naturally as the right one. Look at the page image for each.

Once you have the words, what you do with them is its own subject. The guide to turning a scanned PDF into text covers line breaks, hyphens, page boundaries and special characters in an extracted text file.

Nothing in this post is legal advice. Whether a given document may go to a given tool depends on your duties to the client, any order in the case and your firm's own rules, and those vary by jurisdiction.