Convert PDF to text: 4 methods compared (scanned PDFs included)

You open the PDF, select the text, hit Ctrl+C… and paste a pile of broken symbols. Or worse: you can’t select anything at all, because the document is a photo of a page.

“Convert PDF to text” sounds like a simple task until you discover that some PDFs hand over the text right away and others are nothing but images. The difference between the two cases determines which tool to use — and why some attempts fail silently.

This guide compares 4 methods, shows which one works for each type of PDF, and what to do when the file is scanned.

Why some PDFs give up the text and others don’t

Not every PDF is the same on the inside. In practice, there are 3 types — and it’s the first thing any serious tool checks before extracting:

• Text-based PDF: the content is real text (generated from Word, LaTeX, or Google Docs). Selecting, copying, and searching just works.
• Mixed PDF: part text, part image — common in reports with charts and in partially digitized exams.
• Scanned PDF: every page is a photo. There’s no text in there, just pixels.

Adobe explains the scanned case precisely: when you scan a paper document, the resulting file “contains only image data, not searchable text.” OCR (optical character recognition) is exactly the technology that analyzes the image, recognizes the characters, and creates a text layer — invisible, sitting on top of the photo — that becomes selectable.

There’s a fourth, sneakier case: the PDF that seems to have text, but the layer is corrupted (a botched OCR from an old scanner, for example). You select and paste DASMARAVILHAS in the middle of control characters. This is the case where extracting it “directly” does more harm than good.

The test is simple: try selecting a sentence and pasting it somewhere else. Pasted clean? It’s a text-based PDF. Pasted garbage or nothing? You’ll need OCR.

Method 1 — Copy straight from the PDF reader

Best for: text-based PDFs, short excerpts, one-off use.

If the PDF has a healthy text layer, Ctrl+C does the job:

1. Open the PDF in any reader (Adobe Reader, Chrome, Edge, Preview).
2. Select the passage you want and copy it.
3. Paste it at the destination. Use “Paste without formatting” (Ctrl+Shift+V) to avoid carrying over weird line breaks.

Where it breaks: two-column PDFs (the text order gets scrambled), text with aggressive hyphenation (you end up with exam-ple pasted mid-word), and any scanned page — which simply gives you nothing. For long documents, pasting page by page is guaranteed rework.

Method 2 — Open the PDF in Word

Best for: text-based PDFs you need to edit in full.

Modern Word converts PDFs directly:

1. Open Word and go to File > Open.
2. Select the PDF. Word warns that it will convert a copy — confirm.
3. The document becomes an editable .docx.

Microsoft itself details this method’s limits: recognition accuracy depends on the quality of the original, handwritten text is rarely recognized, and the official recommendation is to review the result after conversion. In other words: it works reasonably well for text-based PDFs with a simple layout, and disappoints on poor-quality scans or complex layouts (tables and columns tend to fall apart).

Method 3 — Google Docs (built-in OCR)

Best for: small scanned PDFs, occasional use, nothing to install.

Google Docs has OCR built into Drive:

1. Upload the PDF to Google Drive.
2. Right-click the file and choose Open with > Google Docs.
3. Google processes the file and opens a new document: the original image at the top and the extracted text below, editable.

For a scanned PDF, this is often the fastest free route. The drawbacks show up with size: large files take a long time or fail, and the layout of complex documents (multiple columns, dense tables) gets scrambled in the conversion. And the result is just raw text — if you need a summary or structure, that’s where another job begins.

Method 4 — Sintesy: the PDF becomes editable text + a synthesis

Best for: large PDFs, scanned ones, and the real-world use case — turning the content into something useful.

Sintesy was built in Brazilian Portuguese first, but the route that matters here is its file import flow:

1. On the dashboard, click New > Upload file — or drag the PDF anywhere on the screen and the import panel opens on its own.
2. The limit is 64 MB per PDF — above that, the warning is clear: split the file into smaller parts and upload them separately.
3. Sintesy classifies the PDF automatically (text, mixed, or scanned) and extracts the content. If the text layer comes in corrupted, it detects the broken characters and reprocesses the page with visual OCR, page by page.
4. The result comes out as a structured synthesis — not a loose block of text. From there you can export to DOCX or PDF to edit wherever you want.

The difference isn’t just the OCR: it’s what happens after it. A 100-page tender document doesn’t become 100 pages of text in your editor — it becomes organized content, with the points that matter. And you can go further with the same file: chatting with the document and turning the PDF into a mind map are two clicks away from the same synthesis.

Scanned PDFs: what actually changes

When the original is an image, extraction quality depends on image quality. Rules that hold for any OCR:

• 300 DPI is the standard for scanning — lower resolutions increase recognition errors.
• A flat page and even lighting make more difference than any tool. A crooked phone photo, with a shadow, delivers text with holes in it.
• A clean image, with no stamp marks or scribbles over the text.

It’s the same logic as audio transcription: if the input is bad, no AI gives you magic back — the problem is almost always the source material. And when OCR makes a mistake, it rarely tells you: Adobe, for example, flags questionable words as “suspect” for manual review — in tools without that marker, the error goes unnoticed until someone rereads the text.

Quick comparison

After extraction: the 3 mistakes that ruin the result

1. Trusting OCR without reviewing it. Proper names, document numbers, amounts, and dates are where every OCR fails first. Always review those fields — in contracts and tender documents, they’re exactly what can’t go wrong.
2. Pasting a PDF table straight into Word. The table turns into run-on text or spreads across broken columns. If the document has lots of tables, choose a method that preserves structure (or rebuild the table from scratch — it’s almost always faster).
3. Ignoring what comes after extraction. Raw text from a scanned PDF comes with hyphenation, weird line breaks, and repeated page headers. If the goal is to study or consult it, the short path is turning the PDF into a mind map or studying the PDF with AI without rereading everything — instead of cleaning up text by hand for an hour.

What about when the PDF contains personal data?

Contracts, medical records, invoices, meeting minutes — most PDFs someone wants to transcribe contain personal data, and sometimes sensitive data (health, biometrics). The LGPD (Brazil’s data protection law) treats these cases more strictly than ordinary data: Article 11 only allows processing sensitive data under specific conditions, with more transparency than the standard.

In practice, this means three simple precautions: prefer tools that don’t redistribute your content to third parties, don’t upload other people’s documents to generic services “just to test,” and delete the file from the service once you’re done. If this is a company workflow, the source document must have some connection to the declared purpose — worth aligning with legal before setting up a recurring flow.

FAQ

Does it work with handwritten scanned PDFs? Rarely. Microsoft itself warns that handwritten text is almost never recognized — OCR was built for printed text. If the document is handwritten, the way to go is transcribing it manually by dictating into a voice tool (which is faster than typing).

Do I have to pay to convert a PDF to text? Not for one-off cases: copying from the reader and Google Docs OCR are free. For large or scanned PDFs, or when you need the content structured, Sintesy’s free plan processes files up to 64 MB with a daily synthesis limit; the Pro plan unlocks volume and DOCX export.

What if the PDF is copy-protected? The selection block is a file permission, not a law of physics — any reader with read permission can “print” the same content into a new PDF. The catch is ethical and legal: if the document has a copy lock, there’s a reason for it. Transcribe only what you have the right to use (quoting a passage with attribution is fair use; reproducing the whole work is not).

Summary: choose by PDF type

• Text-based PDF, quick excerpt → copy from the reader (Ctrl+Shift+V when pasting).
• Text-based PDF, edit everything in Word → File > Open in Word itself.
• Small scanned, one time only → Google Docs via Drive.
• Large, scanned, or when the text is a means and not an end → Sintesy, which extracts with automatic OCR and hands back structured content you can consult in a chat.

And if your case is transcribing meeting or class audio into Word, that’s a different route — see the step-by-step for transcribing audio into Word, which covers the formats Word accepts and the monthly limit of native transcription.