OCR PDF: Text From Scans, On Your Device
Read the text in scanned PDFs with Tesseract OCR in your browser. It finds the script, scores every page, and lets you fix words before you copy or download.
Rate this toolOCR Studio
Several PDFs · 83 languages · confidence per page · .txt or copy
Your scans
Drop scanned PDFs here
or click to browse · up to 100 MB each
Document language
Latin text maps to English, Devanagari to Hindi, and so on.
The first run downloads the OCR engine and each language's data from a public CDN, then your browser keeps it. Your PDF is not sent.
Extracted text
- OCR languages
- 83OCR languages
- Review threshold
- 75%Review threshold
- Largest PDF
- 100 MBLargest PDF
- Files uploaded
- 0Files uploaded
After your OCR
Tools that pick up where this one leaves off.
Read a scanned PDF in three steps
From a scan to text you can copy, search and edit: no account, and the PDF stays on your device.
- STEP 01
Add scanned PDFs
Drop one or more PDFs, up to 100 MB each. They are read one after another.
Read on this device - STEP 02
Detect or pick
Auto-detect reads the script on page 1. Or pick one or more languages yourself.
83 languages - STEP 03
Check and export
Open pages under 75%, fix the marked words, then copy the text or save a .txt.
Edits are kept
How to use this OCR tool
OCR turns a picture of text into real characters. The studio shows how sure it was, page by page and word by word.
- 1
Add the scan
Use the PDF straight from the scanner or scanning app. Each page is rendered and read in your browser.
- 2
Check the language
The detected language shows in the settings. If it's wrong, pick languages and re-run.
- 3
Review low pages
Switch to "To check" to see pages under 75%. Sort by lowest first to start with the worst.
- 4
Fix, then export
Click into a page's text to correct it. Copy all, or download a .txt with a heading per page.
How to get better OCR results
The scan matters more than any setting. These come from Tesseract's own guidance.
Scan at 300 dpi
Tesseract's documentation says it works best on images of at least 300 dpi.
- 300 dpi or more
- Black and white is fine
- Avoid phone zoom
Keep pages straight
Skewed pages make Tesseract split lines badly, which hurts the whole page.
- Lay the page flat
- Square to the camera
- Rotate before OCR
Trim big margins
Very large empty borders around a little text can make Tesseract return an empty page.
- Crop to the text
- Keep a small margin
- Remove black scanner edges
Match the script
A model only knows its own script. Hindi read as English gives nonsense.
- Check the detected language
- Add both for mixed pages
- Re-run if it looks off
83 languages, and how they are picked
Auto-detect finds the writing system. Manual mode lets you combine languages.
Indian languages
English plus Hindi, Bengali, Tamil, Telugu, Marathi, Gujarati, Kannada, Malayalam, Punjabi, Odia, Assamese, Urdu, Nepali, Sanskrit.
World languages
Spanish, French, German, Portuguese, Italian, Russian, Arabic, Chinese, Japanese and Korean, among others.
Auto-detect
Reads the script on page 1 and maps it to one language: Latin to English, Devanagari to Hindi, Han to Chinese.
Several at once
In manual mode, tick two or more. They are read together, which suits pages that mix scripts.
Confidence
Each word gets a score from Tesseract. A page's score is the average. 80% and up shows green.
Marked words
Words under 75% are highlighted with a dotted underline, so you know exactly where to look.
Auto-detect finds the script, not the language. Marathi and Hindi both use Devanagari, so a Marathi scan is read with the Hindi model unless you pick Marathi yourself.
Who uses OCR
The same studio, set up for different paperwork.
Office and admin staff
Pull text out of scanned letters, circulars and records instead of retyping them.
Regional language documents
Land records, notices and certificates in Indian languages, read with the right model.
Students and researchers
Copy a passage from a scanned book chapter or paper, then cite it properly.
Small businesses
Get amounts and names out of scanned bills, then check every number by eye.
What happens to your scan
Straight from how the tool is built.
Read on your device
pdf.js renders each page at about 144 pixels per inch, and Tesseract.js reads it.
- No PDF sent to a server
- Works on private papers
- One document at a time
Engine loads once
The first run fetches the OCR engine and the language data from the jsDelivr CDN.
- Only the engine comes in
- Language data is cached
- Your pages stay here
Kept if you refresh
The extracted text, with your edits, is saved in this browser.
- Text only, no PDF
- Remove a document to clear it
- Add the PDF again to re-run
Password-protected PDFs can't be read. Remove the password first with the PDF Password Remover, then add the file again.
Working with OCR text
Habits that save you from errors later.
- 1
Keep the scan
The scan is the record. OCR text is a copy that can contain mistakes.
- 2
Check numbers by eye
Amounts, dates and ID numbers are where a single wrong character matters most.
- 3
Name the export
The .txt is named after the PDF with _ocr added, so the two stay paired.
- 4
Clear when done
On a shared computer, remove the documents so their text is cleared from the browser.
Auto-detect or pick languages?
Both use the same engine. The difference is how the language model is chosen.
| What you get | Auto-detect | Pick languages |
|---|---|---|
| Setup needed | ✓ None | Choose from the list |
| Pages looked at to choose | Page 1 only | ✓ Your choice applies to all |
| Mixed scripts on a page | ✗ One language | ✓ Tick several |
| Same-script languages (Hindi, Marathi) | Picks one model per script | ✓ Exact language |
| Extra download | Detection data too | ✓ Only your languages |
If detection fails, the tool falls back to English. Check the "Detected" line in the settings after each run.
Pro tips for cleaner text
Small things that save time when you check the result.
Lowest first
Sort by confidence and fix the worst pages first.
Compare
Open a page and press Compare to see the scan next to its text.
Follow the marks
Dotted words are the unsure ones. Start there.
Mixed pages
Tick both languages, such as English and Hindi, and re-run.
One page only
Copy page grabs just that page's text, with your edits.
Sideways pages
Turn them upright first with Rotate PDF.
Big margins
Trim wide borders with Crop PDF before OCR.
Need Word?
For a .docx, use PDF to Word, which also runs OCR.
Confidence is Tesseract's own estimate. A page can score well and still have a wrong digit, so always read numbers you will rely on.
OCR PDF or PDF to Word?
Both use Tesseract.js in your browser. They suit different jobs.
OCR PDF
Reads every page of several PDFs, scores them, marks unsure words and lets you fix the text. Exports .txt or the clipboard.
Best when the whole file is scanned
PDF to Word
Reads a PDF's own text with headings, and runs OCR only on pages that need it. Exports a Word document.
Best for mixed text and scanned pages
What people use OCR for
The same studio, six everyday jobs.
Old letters and forms
Reuse wording from paperwork that only exists as a scan.
Regional records
Read land records and notices in Indian scripts.
Book and paper extracts
Quote a passage without retyping it.
Bills and receipts
Pull names and amounts out to check or file.
Quick copy
Grab one paragraph from a scanned page.
Private papers
Medical and bank documents, read without uploading.
Scan too heavy to share afterwards? Shrink it with Compress PDF.
Frequently asked questions
Quick answers about languages, accuracy, privacy and exports.
Which languages are supported?
How accurate is the OCR?
How long does OCR take?
What do I get at the end?
Can it read password-protected PDFs?
What people say
Reviews from people who used this tool.
Loading reviews…
Popular presets
Jump to a version of this tool already set up for a specific need.
- PresetOCR PDF for Hindi (Devanagari)OCR PDF Hindi: pull Devanagari text from scanned bank statements, exam papers and govt notices.Open preset
- PresetScanned PDF to Text ConverterConvert scanned PDFs to plain text with OCR in your browser.Open preset
- PresetOCR PDF to Editable TextTurn a scanned PDF into editable text with OCR in your browser.Open preset
- PresetOCR PDF for Marathi (Devanagari)OCR PDF for Marathi: pull Devanagari text from scanned 7/12 extracts, MSEB bills and MPSC papers so you can copy and edit it.Open preset
- PresetOCR PDF for Gujarati (Gujarati)Gujarati OCR for PDF: extract real Gujarati text from scanned 7/12 records, GSEB papers and PGVCL bills to copy and search.Open preset
- PresetOCR PDF for Tamil (Tamil)OCR PDF Tamil: extract text from scanned patta chitta, TNPSC papers and EB bills to search and copy it.Open preset
- PresetOCR PDF for Telugu (Telugu)OCR PDF Telugu: pull Telugu text from scanned pattadar passbooks, Dharani printouts and TSPSC or APPSC papers so it becomes searchable.Open preset
- PresetOCR PDF for Bengali (Bengali)OCR PDF for Bengali: extract text from scanned porcha and khatian records, WBBSE papers and KMC notices as editable Bengali text.Open preset
- PresetOCR PDF for Kannada (Kannada)Kannada OCR for PDF: copyable text from scanned RTC and pahani extracts, KPSC papers and BESCOM bills.Open preset
- PresetOCR PDF for Malayalam (Malayalam)OCR PDF Malayalam: extract Malayalam text from scanned PDFs like Kerala PSC papers, KSEB bills and village office certificates.Open preset
- PresetOCR PDF for Punjabi (Gurmukhi)OCR PDF Punjabi: read Gurmukhi text from scanned jamabandi records, PSEB papers and PSPCL bills, then copy or search it.Open preset
- PresetOCR PDF for Odia (Odia)OCR PDF Odia: extract Odia text from scanned Bhulekh ROR records, OSSC papers and BSE Odisha mark sheets with an Odia OCR model.Open preset
- PresetOCR PDF for Urdu (Nastaliq)OCR PDF for Urdu: read Nastaliq text from scanned certificates, newspapers and government notices, kept in right to left order.Open preset
More free PDF tools
Free, with no sign-up, just like this one.
- Free toolPDF to WordConvert PDF to editable Word with OCR, font detection, and page preview - 100% free, no signupTry it now
- Free toolWatermark PDFAdd professional text or image watermarks to PDF documents.Try it now
- Free toolSign PDFDraw, type, or upload a signature; save templates; drop date stamps and initials - sign any PDF entirely in your browser.Try it now
- Free toolWord to PDFConvert one or many .docx files to PDF - pick page size and margins, batch-download as a ZIP, all in your browser.Try it now
- Free toolJPG to PDFCombine JPG, PNG, WebP, and other images into a single, clean PDF - page size, orientation, and margins of your choice.Try it now
- Free toolPDF to PowerPointConvert PDF files to real .pptx slides: each page is rendered as a high-fidelity image so layout is preserved exactly (text isn't separately editable; use PDF to Word for that).Try it now