How to use
- Add scanned PDFs or images.
- Choose the document language and output (text or searchable PDF).
- Review the recognized text and confidence per page, then download.
Worked example
A one-page 300 DPI scanned invoice reading “INVOICE 2026 TOTAL 42.50” was recognized in about 8 seconds on a laptop, including loading the engine, and saved as a searchable PDF whose text can be selected, searched and copied.
Supported formats and limits
| Input | PDF, PNG, JPEG, WebP, TIFF, BMP |
|---|---|
| Output | TXT, Searchable PDF |
| Limits | Up to 100 pages per run. The first run downloads about 7 MB (recognition engine and English data) from this site. |
| Engine | tesseract.js 7 (LSTM) in a Web Worker; pdf.js page rendering at 300 DPI; pdf-lib text layer |
Limitations
- Handwriting is not recognized reliably.
- Accuracy drops on low-resolution (under 200 DPI), skewed or low-contrast scans.
- Only English recognition data is included at the moment.
- Multi-page TIFF files: only the first page is read. Image pages are sized assuming a 300 DPI scan (150 DPI for small images).
Questions
Which languages are supported?
English only at the moment. Text in other languages will be recognized poorly.
What does a searchable PDF look like?
The original page images stay the same, with an invisible text layer on top that you can select, search and copy. Pages that already have text can be skipped so they do not get a second layer.
How accurate is it?
Clear printed text at 200 DPI or more usually reads well. Accuracy drops on low-resolution, skewed or low-contrast scans, and handwriting is not recognized reliably, so check important values.
Guides
Privacy
Runs on your device. Files and text are processed in this browser tab and are not uploaded.
See the privacy policy for how toolsdocks handles data.