Fes que el text d'un PDF es pugui cercar i copiar.
Com funciona:
Fes clic per seleccionar un fitxer o arrossega'l i deixa'l anar
Els teus fitxers no surten mai del dispositiu.
Seleccionats: Cap
Només es reconeixeran aquests caràcters. Deixa el camp buit per admetre'ls tots.
Només es reconeixeran aquests caràcters. Deixa el camp buit per admetre'ls tots.
S'està inicialitzant...
El PDF amb text cercable està a punt. També pots copiar o descarregar el text extret que apareix a continuació.
S'està processant...
Fes clic o arrossega i deixa anar el fitxer per començar
Fes clic al botó de processament per començar
Desa el fitxer processat a l'instant
Open the file in any viewer and try to select a line of text. If nothing highlights or search finds no matches, the pages are just images (a scan or a photo) and OCR is what adds the missing text. A PDF that already has selectable text doesn't need this tool.
Over 100 Tesseract language packs are listed, from English and Spanish to Arabic, Hindi, Japanese, and Chinese. Tick every language that appears in the document; the packs are loaded together so mixed-language pages are recognized in one pass. Some self-hosted deployments bundle a shorter list and say so above the language picker.
No. The original pages are copied unchanged and an invisible text layer is drawn on top, positioned word by word to match the scan. You see the same image as before, but you can now search, select, and copy the text.
High (288 DPI) is the default and handles typical office documents. Drop to Standard (192 DPI) when speed matters or the file is long, and use Ultra (384 DPI) for small print, footnotes, or poor scans. Higher settings take noticeably longer since every page is rendered at that size before recognition.
It restricts the output to a fixed set of characters, which cuts down on errors like a lowercase l being read as a 1. The Invoice preset allows only digits and the symbols $ . , / - # :, Numbers Only suits serial numbers or meter readings, and Custom lets you type your own set.
Not reliably. Tesseract is trained on printed text, so neat block capitals sometimes work but cursive mostly produces noise. Results are best on clean, well-lit printed pages; the Binarize option helps with faded or low-contrast scans, and Deskew PDF straightens tilted ones first.
Yes. When OCR finishes, the extracted text appears in a box where you can review it, copy it to the clipboard, or download it as ocr-text.txt. The searchable PDF download is separate, so you can take either or both.
No. Tesseract runs as WebAssembly inside your browser; the only downloads are the OCR engine and the language packs you selected, fetched the first time they're needed. The scanned pages and the recognized text stay on your device.