📄 PDF to Text
Extract text from any PDF. Download as plain .txt or Word .docx. Everything runs locally — no upload.
Scanned PDF with no text layer? An optional built-in OCR pass can read the pages as images — you'll be offered it only if it's needed.
🔒 Your documents are never uploaded to any server — all processing happens in your browser. Library files are served locally by this server.
About PDF to Text
Got a PDF full of text you need to copy, search, edit, or process? Instead of painstakingly copying it paragraph by paragraph, this tool extracts all the text content from any PDF in one go and lets you download it as a plain .txt file or a .docx Word document — ready to open, edit, and reformat in any text editor or word processor.
Everything runs locally in your browser — your PDF is never uploaded to any server. Useful for extracting text from contracts, research papers, reports, ebooks, and scanned documents that have a text layer.
Purely scanned PDFs — photocopies, phone scans, faxes — carry no embedded text layer, so a normal extraction returns empty pages. When that happens this tool detects it and offers an optional OCR pass that reads the pages as images instead, using the open-source Tesseract engine in 17 languages including Vietnamese, Chinese, Japanese, Korean, Thai and Arabic. OCR never starts on its own: you pick the language and the page range and press the button, and the engine is only downloaded at that moment, so ordinary text PDFs stay as fast as they were. Recognised text drops straight into the same page list and the same .txt / .docx downloads.
Vietnamese PDFs that extract as gibberish — TiÕng ViÖt or Tieáng Vieät instead of Tiếng Việt — are not damaged. They were written in a pre-Unicode Vietnamese font (TCVN3/ABC .VnTime, VNI-Windows or VISCII), and the characters you are seeing are that font's byte values. This tool recognises all three automatically and converts them back to real Unicode in one click, and it re-composes accents (NFC) on every document even when no conversion is needed, so Ctrl+F, sorting and Word behave. The original extraction is always kept, so you can undo or try a different encoding at any time. VIQR and Telex — the conventions for typing Vietnamese on an ASCII keyboard — can also be converted, but must be chosen manually. A document that mixes two legacy fonts — common when a section was pasted in from another file — is handled per font rather than as a whole, because converting all of it as one encoding would repair most of it and quietly damage the rest.
OCR proofreading. Vietnamese OCR errors are usually real words, not nonsense: cái misread as cói, đã as đỡ — the tone mark survives and only the vowel changes, so no spell checker can see it. After an OCR pass this tool compares the recognised words against the document's own vocabulary: a word that appears once, one confusable step away from a word the same document uses forty times, is almost certainly a misreading of it. Suggestions are listed for you to read and applied only on request, they never touch text that came from the PDF's own text layer (which is exact), and nothing is sent anywhere — the judgement is made from your file alone, in your browser.
PDF is read locally — never uploaded