How to Convert a Scanned PDF to Searchable Text (OCR) — Free
You can make any scanned PDF searchable in seconds with OCR — upload it to a free online tool and get back a PDF with selectable, copyable text. This guide covers the fastest method plus offline options, accuracy tips, and batch processing.
Sarah Chen·February 19, 2026·13 min read
The Problem with Scanned PDFs
A scanned PDF is essentially a collection of photographs. When you scan a paper document, the scanner captures an image of each page -- not the text on it. The result is a PDF file where every page is a flat image. You cannot search for words, select and copy text, or use screen readers for accessibility. The text is visible to your eyes but invisible to your computer.
This creates real problems. You cannot find a specific clause in a 200-page scanned contract by pressing Ctrl+F. You cannot copy an address or phone number from a scanned letter. You cannot extract data from a scanned invoice into a spreadsheet. And for anyone using assistive technology, scanned PDFs are completely inaccessible.
OCR (Optical Character Recognition) solves this by analyzing the images in a scanned PDF, recognizing the text characters, and adding an invisible text layer on top of each page. The result is a searchable PDF -- visually identical to the original scan but with fully selectable, searchable, copyable text underneath.
Before and after comparison of a scanned PDF showing non-selectable text versus OCR-processed searchable text
Try these conversions
Free, in your browser — no signup, files auto-delete in 2 hours.
Modern OCR engines use a multi-stage process to convert images to text:
Preprocessing -- The image is cleaned up: deskewed (straightened), denoised (speckles removed), binarized (converted to black and white for better contrast), and scaled to an optimal resolution for recognition.
Layout analysis -- The engine identifies text regions, separating them from images, tables, headers, footers, and margins. It determines reading order, column structure, and paragraph boundaries.
Character recognition -- Each text region is analyzed character by character. Modern engines like Tesseract 5 use LSTM (Long Short-Term Memory) neural networks trained on millions of text samples. The network outputs a probability for each possible character and selects the most likely sequence.
Post-processing -- The recognized text is checked against language dictionaries and grammar rules to correct likely errors. For example, "rn" misrecognized as "m" (a common OCR error) can be corrected by checking if the resulting word exists in the dictionary.
Text layer creation -- The final recognized text is placed into an invisible layer that precisely overlays the original image. Each word is positioned to match its visual location on the page.
Method 1: Convert Scanned PDFs Online
Our online OCR tool processes scanned PDFs entirely in your browser with no software to install.
Upload your scanned PDF (supports files up to 50 MB)
Select the document language (or multiple languages for multilingual documents)
Choose the output type:
Searchable PDF -- Original images with invisible text layer (most common)
Text-only PDF -- Recognized text reformatted as a text document
Text file (.txt) -- Plain text extraction only
Click Process and wait for OCR to complete
Download your searchable PDF
Our OCR engine supports 17 languages including English, Spanish, French, German, Italian, Portuguese, Dutch, Russian, Chinese, Japanese, Korean, Arabic, Hindi, Turkish, Polish, Swedish, and Norwegian.
What You Get
The output is a PDF that looks exactly like your original scan but with a hidden text layer that enables:
Ctrl+F search -- Find any word or phrase instantly
Text selection -- Click and drag to select text, then copy to clipboard
Text-to-speech -- Screen readers can read the document aloud
Data extraction -- Copy text into other documents, spreadsheets, or databases
Method 2: Adobe Acrobat OCR
Adobe Acrobat Pro includes powerful OCR functionality:
Open the scanned PDF in Acrobat Pro
Click Scan & OCR in the right panel
Click Recognize Text > In This File
Set language and output settings
Click Recognize Text
Acrobat's OCR is highly accurate and handles complex layouts well, but requires a paid subscription ($22.99/month).
Method 3: Free Desktop Tools
Tesseract OCR (Open Source)
Tesseract is the most widely used open-source OCR engine, originally developed by HP and now maintained by Google. It powers many commercial OCR products.
# Install Tesseract
# macOS: brew install tesseract
# Ubuntu: sudo apt install tesseract-ocr
# Convert scanned PDF to searchable PDF
# First extract images, then OCR each page
pdftoppm -r 300 scanned.pdf page
tesseract page-1.ppm output-1 pdf
For multi-page PDFs, tools like ocrmypdf wrap Tesseract with convenient batch processing:
# Install ocrmypdf
pip install ocrmypdf
# One command to OCR an entire PDF
ocrmypdf input.pdf output.pdf --language eng --deskew --clean
NAPS2 (Windows, Free)
NAPS2 (Not Another PDF Scanner 2) is a free Windows application that combines scanning and OCR:
Open NAPS2
Import your scanned PDF (File > Import)
Click OCR and select the language
Export as searchable PDF
Optimizing Scan Quality for Better OCR
The accuracy of OCR depends heavily on the quality of the input scan. A clean, high-resolution scan can achieve 99%+ accuracy, while a poor scan might drop to 70-80%.
Factor
Optimal Setting
Impact on Accuracy
Common Mistakes
Resolution (DPI)
300 DPI (600 for small text)
Critical -- below 200 DPI accuracy drops sharply
Scanning at 72-150 DPI for smaller files
Color mode
Grayscale (not color, not B&W)
High -- grayscale preserves edge detail
Using pure B&W threshold which destroys antialiasing
Contrast
High contrast between text and background
High -- low contrast causes character confusion
Scanning faded documents without contrast adjustment
Skew angle
0 degrees (perfectly straight)
Medium -- most OCR engines auto-deskew
Placing pages crooked on the scanner bed
Page cleanliness
No stains, marks, or folds across text
Medium -- noise can be mistaken for characters
Scanning coffee-stained or creased pages
Font size
10 pt or larger
Medium -- very small text needs higher DPI
Footnotes and fine print at low resolution
Font type
Standard printed fonts (serif or sans-serif)
Low-Medium -- unusual fonts reduce accuracy
Decorative, script, or damaged typeface
Pro Tip: If you have the option to re-scan, always choose 300 DPI in grayscale. Color scans are 3x larger than grayscale without improving OCR accuracy (the engine converts to grayscale internally anyway). The extra file size slows processing and storage without any recognition benefit. Reserve 600 DPI for documents with very small text (6-8 pt) like footnotes, legal fine print, or technical specifications.
Handling Different Document Types
Standard Business Documents
Letters, memos, reports, and contracts with standard fonts and clean layouts convert with 95-99% accuracy. These are the easiest documents to OCR.
Workflow: Scan at 300 DPI grayscale, run OCR with English language, verify a few key sections.
Financial Documents
Bank statements, invoices, and receipts contain numbers, currency symbols, and tabular data. OCR accuracy for numbers is critical because a single wrong digit can have real consequences.
Workflow: Scan at 300 DPI, run OCR, then verify all numeric values against the original. For extracting table data into spreadsheets after OCR, see our guide on how to convert PDF to Excel.
Historical and Degraded Documents
Old documents with yellowed paper, faded ink, and aged typefaces present the greatest OCR challenge. The text may be partially illegible even to human readers.
Workflow: Scan at 600 DPI, preprocess to enhance contrast, use OCR with post-correction. Accept that accuracy may be 80-90% and plan for manual review.
Multilingual Documents
Documents containing text in multiple languages (common in international contracts, academic papers, and government documents) require multi-language OCR.
Workflow: Select all relevant languages in the OCR settings. Most engines support specifying 2-3 languages simultaneously. Accuracy may be slightly lower than single-language processing because the dictionary lookup has more candidates.
Handwritten Documents
Handwriting recognition is significantly harder than printed text recognition. Modern OCR engines can handle neat, consistent handwriting with moderate accuracy (70-85%) but struggle with cursive, messy, or inconsistent handwriting.
Workflow: Use an OCR engine with handwriting support (not all engines include this). Expect extensive manual correction. For critical documents, manual transcription may be more efficient.
Different document types showing varying OCR accuracy levels
Post-OCR Quality Assurance
OCR is not perfect. Even the best engines make mistakes, and uncorrected OCR errors can cause problems -- especially in legal, financial, and medical documents. Here is a systematic quality assurance process.
Common OCR Errors
Error Type
Example
Cause
Detection Method
Character confusion
"rn" read as "m", "l" read as "1"
Visual similarity between characters
Spell-check, manual review
Merged words
"the report" read as "thereport"
Tight character spacing
Spell-check highlights unknown words
Split words
"information" read as "infor mation"
Worn or uneven ink, damaged print
Search for unusual spaces
Wrong numbers
"5" read as "6", "0" read as "O"
Similar shapes, low resolution
Manual verification of critical numbers
Missing characters
"document" read as "docment"
Faded ink, low contrast
Spell-check, word count comparison
Extra characters
Specks recognized as periods or commas
Paper noise, scanner artifacts
Manual review of punctuation
Verification Steps
Search for common problem words -- Search the OCR'd text for known error-prone words in your document type
Compare page counts -- Ensure the text layer has content on every page
Spot-check paragraphs -- Read 2-3 random paragraphs against the original image
Verify critical data -- Manually confirm names, dates, amounts, and identifiers
Run spell-check -- Open the searchable PDF in Word or a text editor and run spell-check to catch recognition errors
Batch Processing Scanned PDFs
Organizations with large archives of scanned documents need batch OCR capabilities.
Online Batch OCR
Upload multiple scanned PDFs to our PDF OCR tool simultaneously. Each file is processed independently, and you can download all results as a ZIP archive.
Automated Pipeline
For ongoing document digitization:
# Watch a folder for new scanned PDFs and OCR them automatically
inotifywait -m /incoming -e create -e moved_to |
while read dir action file; do
ocrmypdf "/incoming/$file" "/processed/$file" \
--language eng --deskew --clean --optimize 1
done
Pro Tip: When batch processing hundreds of scanned PDFs, prioritize by business need rather than processing everything at once. OCR the documents people search for most frequently first (contracts, policies, financial records), then work through the archive chronologically. This delivers immediate value while the full archive is still being processed.
OCR and PDF Accessibility
Making scanned PDFs searchable is a critical step toward document accessibility, but it is not the complete solution. A truly accessible PDF also needs:
Document structure tags -- Headings, paragraphs, lists, and tables tagged for screen reader navigation
Reading order -- Content ordered logically, not just visually
Alt text for images -- Descriptions for non-text visual elements
Language specification -- Document language declared for correct pronunciation
OCR creates the text layer that makes other accessibility improvements possible. Without OCR, a scanned PDF is a complete barrier to anyone using a screen reader. With OCR, you have a foundation to build on.
For a comprehensive guide to PDF accessibility requirements and implementation, see our PDF accessibility guide.
Combining OCR with Other PDF Operations
OCR is often one step in a larger document workflow:
For extracting specific pages from a large scanned document before OCR processing, our extract pages tool lets you pull out only the pages you need, reducing OCR processing time.
Workflow diagram showing scanned document going through OCR then into various output formats
OCR Output Formats
After OCR, you may want the content in a format other than searchable PDF:
Searchable PDF -- The standard output; original images with invisible text layer
OCR processing time depends on the number of pages, scan resolution, document complexity, and the OCR engine used:
Simple documents (1-10 pages): 5-30 seconds
Medium documents (10-50 pages): 30 seconds to 3 minutes
Large documents (50-200 pages): 3-10 minutes
Archive batches (200+ pages): 10-30 minutes
Higher resolution scans take longer to process but produce better results. The processing time investment is almost always worth the accuracy improvement.
Our online tool processes files for free up to the size limit, with no page count restriction. For high-volume processing, our Pro plan offers faster processing, larger file sizes, and API access for automated workflows.
Best Practices Summary
Scan at 300 DPI minimum -- Resolution is the single biggest factor in OCR accuracy
Use grayscale, not color -- Same accuracy, 3x smaller files, faster processing
Deskew before OCR -- Straighten crooked scans for better recognition
Select the correct language -- The dictionary and character set must match the document
Verify critical data manually -- Never trust OCR blindly for numbers, names, or legal text
Keep the original scans -- Always retain the source images in case you need to re-OCR with better settings
Process at the right time -- OCR before merging, reordering, or other PDF operations
Compress after OCR -- The text layer adds file size; compression rebalances it
Converting scanned PDFs to searchable documents transforms dead images into living, usable text. Whether you are digitizing a paper archive, processing incoming scanned mail, or making documents accessible, OCR is the essential technology that bridges the gap between physical paper and digital searchability.
For deeper OCR guidance including language-specific tips and troubleshooting, see our comprehensive tutorial on how to OCR scanned documents.
Frequently Asked Questions
Vanliga fragor
Upload the scanned PDF to a free online OCR tool such as ConvertIntoMP4's PDF OCR tool. Select the document language, choose 'Searchable PDF' as the output type, and click Process. The tool adds an invisible text layer over each page — you can then Ctrl+F search, select text, and copy content from the result. No software installation required.
OCR (Optical Character Recognition) analyzes the images in a scanned PDF and recognizes the text characters. It goes through five stages: preprocessing (deskew, denoise, binarize), layout analysis (find text regions), character recognition (LSTM neural network matches characters), post-processing (dictionary checks), and text layer creation (invisible text placed over each original page image).
Scan at 300 DPI minimum using grayscale mode. Below 200 DPI, OCR accuracy drops sharply. Color scans are three times larger than grayscale without improving recognition because the engine converts to grayscale internally anyway. For very small text (footnotes, legal fine print under 8 pt), use 600 DPI.
Yes. The most common output — Searchable PDF — keeps the original scanned images exactly as they are and adds an invisible text layer underneath. The document looks identical to the original scan but now supports text search, selection, copy-paste, and screen readers.
Yes. Upload multiple scanned PDFs simultaneously to the online OCR tool and download all results as a ZIP archive. For automated pipelines, the open-source tool ocrmypdf (wrapping Tesseract) can watch a folder and OCR incoming files automatically.
Common OCR errors include character confusion ('rn' misread as 'm'), merged or split words from tight spacing, and wrong numbers ('5' misread as '6'). Low scan resolution, poor contrast, and faded ink are the main causes. Scan at 300 DPI in grayscale, run spell-check on the output, and manually verify critical numbers, names, and dates.
