Fix PDF text that copies as symbols

First check which problem you have

Search for a visible word and paste one sentence into a plain-text editor. No selectable words usually means an image-only scan. Selectable words that paste as unrelated symbols can indicate a missing or incorrect font-to-Unicode mapping. An explicit copy restriction is a permissions issue. These problems need different checks.

Try ordinary extraction before rebuilding pages

For selectable text, test PDF to Text. If extraction is readable, there is no need to rasterize the document. For image-only scans or mixed documents, automatic OCR can add recognition where needed. Check the downloaded result rather than reopening the original.

Use Forced OCR when the existing text is unusable

1. Keep the original PDF and open OCR PDF.
2. Enable Forced OCR. This applies to the whole document: every page is rendered into an image before a new text layer is recognized.
3. Select the printed languages and a suitable quality preset. Use Standard as a starting point; compare High quality for difficult print.
4. Download the result, search known phrases and paste sample paragraphs into a text editor.

For a plain-text file, follow the scanned PDF to text workflow after recognition.

Understand what forced recognition changes

Image-first processing replaces the original page content with a rendered appearance and fresh recognized text. Vector detail and interactive forms are flattened. Rewriting does not preserve digital-signature validity or certify accessibility. Retain the original for editing, archival or signed-document requirements.

Deskew can straighten tilted scans. Clean pages can reduce recognition noise. Compare a representative page because preprocessing may affect faint marks. Neither setting restores missing letters.

Handle passwords and permissions separately

Process only documents you are authorized to modify. An opening password is still required when a PDF cannot be opened without it. Unlock PDF is an authorized workflow, not password recovery. OCR is not a way to recover an unknown opening password.

Verify more than one paragraph

Check clear text, small print, tables and pages near the end. OCR can make new mistakes even when copying starts working. Review critical numbers manually. If a page remains unreadable, try a clearer source or an excerpt instead of repeatedly processing a large file.

Service limits and privacy

The hosted tools accept up to 100 MB per file and 500 pages per job, subject to shared usage limits. Selected files are uploaded over HTTPS. Results are temporary for at most 24 hours and can disappear sooner after restart. Download promptly. See document handling, security limits and the large-PDF OCR guide.