Extract text from images — free

Upload your PDF and the tool detects whether it contains a real text layer or just scanned images. Text-layer PDFs export instantly. Image-only PDFs (scanned books, photographed receipts, old reports) run through OCR automatically. Either way you get a clean .txt file with paragraphs, line breaks and section spacing preserved.

Yes. If the PDF is image-only (common for scans, photographed pages or fax exports), the OCR engine kicks in automatically — you do not need a separate tool. Multi-language support is built in, so even bilingual or non-Latin scripts (Chinese, Arabic, Hindi) get extracted in the same pass.

Paragraph breaks, blank lines between sections, bullet point markers and numbered list prefixes are preserved as plain text. Headings come through in upper-case or as their original casing depending on the source font. Visual emphasis (bold, italic) is not encoded in plain text — for that, use the PDF to Word converter instead.

Yes. Enter the password in the prompt after upload, and the tool unlocks the file in memory just long enough to extract the text. The password is never stored on disk or transmitted to third-party services. Locked PDFs without the password cannot be processed — for security reasons, no password-cracking is performed.

Files up to 50 MB and 500 pages process without trouble. Larger documents work too but take longer — a 2000-page legal archive may take a few minutes for OCR. For massive batches, split the PDF first with the PDF Split tool and process each chunk in parallel.

Processing happens on secure servers and files are deleted within minutes. The .txt output is yours — no watermark, no attribution, no tracking. Researchers, journalists, lawyers and students use the tool to extract text from confidential reports knowing the source PDF is not retained beyond the conversion.

Yes. Open the Engine panel, choose your document's language and OCR engine, and the page is read in that script — over 100 languages are supported, including non-Latin and right-to-left ones. If the first pass misreads accented or non-English characters, switch the language and tap Re-run extraction.

Formatted view keeps the page's original layout — columns, spacing and line positions — which helps with tables and receipts. Plain view gives clean, reflowed text that's easier to paste into a document or chatbot. Toggle between them, then copy the text or download it as a .txt file.

This tool hands you the raw text to copy or save as .txt. If you'd rather keep the original PDF but make it Ctrl+F searchable, run it through the Image to Searchable PDF tool — it adds an invisible OCR text layer over the scan so the page looks identical while the words become selectable.

Start with Default and your document's language for common Latin-script languages — it's fast and accurate for everyday text. If the output looks garbled or the script is non-Latin (Arabic, Hindi, Chinese, Cyrillic), switch to Engine 1 or Engine 2, select the matching language from the picker, and tap Re-run extraction — different engines are tuned for different scripts, so trying both takes seconds.

The output box is fully editable, so you can quickly clean up an OCR mistake or trim a section right on screen. Copy to clipboard always copies exactly what's currently in the box, edits included — but Download .txt saves the original file produced by the last Engine/Language run, not your on-screen edits. To keep a correction, use Copy and paste it into your own .txt file, or if the mistake is systematic, switch the Language or Engine and tap Re-run extraction instead of hand-editing.