How to OCR a PDF in Chrome

Chrome's support for modern web APIs — including the File API, Web Workers, Canvas, and WebAssembly — makes it an excellent environment for running Tesseract.js. The OCR engine loads as a JavaScript module, uses a Web Worker thread to avoid freezing the browser UI, and processes each PDF page as a Canvas image. The entire pipeline from file selection to searchable PDF download happens without any network requests after the initial page load. Here's the step-by-step process.

1. 1Open Chrome on any operating system and navigate to lazy-pdf.com/en/ocr
2. 2Click 'Choose File' or drag your scanned PDF from File Explorer, Finder, or your Linux file manager directly into the drop zone
3. 3Select the language of your document from the language dropdown — this loads the correct Tesseract language model for improved accuracy
4. 4Click 'Run OCR', watch the progress bar advance as each page is processed, then click 'Download PDF' to save the searchable version

Tesseract.js runs inside Chrome's V8 JavaScript engine as a Web Worker — a background thread that processes data without blocking the browser's main UI thread. Your PDF file is read from disk using the File API, each page is decoded using pdfjs-dist (a local PDF renderer), rasterized to a Canvas element, and passed to Tesseract.js for character recognition. The resulting text strings are then embedded into the PDF using pdf-lib. Every step happens in Chrome's sandboxed process on your machine. There are no API calls, no WebSocket connections to OCR servers — you can confirm this by opening Chrome DevTools (F12), going to the Network tab, and watching zero requests after the initial page load.

Tesseract.js supports over 100 languages, and LazyPDF's OCR tool lets you select your document's language before processing. The language selection loads a specific trained data model for Tesseract — for example, selecting 'French' loads a model trained on French text patterns, significantly improving accuracy for French characters, accented letters, and common French word shapes. For documents that mix two languages (such as a bilingual contract), choose the primary language. If you frequently process documents in a specific language, Chrome will cache the language model after the first use, making subsequent runs faster. Common language models like English are smaller and load quickly; scripts like Chinese or Arabic have larger model files.

Many people try browser extensions or upload-based online services for PDF OCR. Extensions like Adobe Acrobat extension or Smallpdf upload your document to their servers — which means your file leaves your computer and is processed on hardware you don't control. LazyPDF's in-browser OCR is different: Tesseract.js runs locally in Chrome using your CPU. For a typical 5-page scanned document on a modern laptop, local processing takes 1-3 minutes — similar to many server-based services, but without the privacy tradeoff. Chrome's multi-core JavaScript execution (via Web Workers) also means the OCR process doesn't make Chrome unresponsive; you can browse other tabs while it runs.

After downloading the OCR-processed PDF, you can immediately verify the results in Chrome's built-in PDF viewer. Drag the downloaded file back into a new Chrome tab (or open it from your Downloads bar). Press Ctrl+F (Cmd+F on Mac) to open the search box and type a word you know appears in the document. If OCR worked correctly, Chrome will highlight the matching text. You can also select text with your mouse and copy it with Ctrl+C. If Chrome's search finds no matches, the page may have very low scan quality — try rescanning at higher resolution or with better lighting, then run OCR again.