Convert · Runs on our server
Have a scanned PDF with text you can't search or select? This tool recognizes the text in your scans and adds a selectable layer, so you can copy, search, and edit. It runs on our servers, keeps your files private, and cleans up after itself.
Runs on our server This tool needs a document engine, so the file is uploaded to a private bucket, processed and deleted automatically.
or drop here — PDF up to 50 MB
03 / HOW TO MAKE A SCAN
Choose your PDF file (one file per job, up to 50 MB, max 500 pages) and the tool scans the uploaded file. Under the language settings, tick English, German, French, or any combination - only these three are available. In the resolution field (DPI), set a value from 150 to 400; the default is 300.
That's all there is to it. No need to pick a text layer option; the tool adds recognition automatically to pages that need it.
04 / WHAT THE TOOL DOES
After upload, the file is stored in private object storage and is deleted after 1 hour if you're not signed in, or after 7 days if you are signed in. The file is scanned for malware and its type is verified from its actual content - not its name - so a file that isn't truly a PDF is rejected before any work.
Each page is checked: if it already contains selectable text (about a dozen real characters), that page is passed through unchanged, preserving its original sharpness. Only the pages that truly need OCR are re-imaged at your chosen DPI and run through Tesseract OCR.
The chosen languages are verified before the job starts, so if one isn't installed you'll know immediately. After processing, the tool re-opens its output and reads it back to make sure the page count matches and the text is extractable; if not, the job fails rather than handing you a file that only looks searchable.
05 / LIMITS AND WHAT TH
Recognition is not perfect. Handwriting, unusual typefaces, low-resolution scans, and touching columns will reduce accuracy. No OCR software can fix these issues. If your source is a clean print scan at decent resolution, you'll have the best chance of good results.
The tool only handles one PDF per job, and a maximum of 500 pages per file. Larger batches need to be split. Also, only English, German, and French are supported - if your scan is in another language, it won't be recognized.
06 / QUALITY NOTES: DPI
The chosen DPI affects how well the OCR works. For most printed text, 300 DPI is a solid default. Lower settings (near 150) may miss details, while higher settings (up to 400) can improve recognition on small or faint text but increase processing time and file size.
Because digital pages with existing text are copied through untouched at their original resolution, you won't lose quality on those pages. Only the scanned pages are processed, so a mixed document keeps its clear digital pages while gaining searchable text on the scanned ones.
07 / AFTER THE JOB: WHA
While the job runs, the page shows uploading, then queued, then running with a percentage. Once done, your download is ready.
If the tool added a recognized text layer, you'll see the warning OCR_TEXT_LAYER_ADDED. That's the only warning it gives, and it's normal - it means recognition was added, but it might not be perfect. If you need a flawless result, review the output carefully, especially if the source was low quality.
FAQ
No. Pages that already have selectable text are left untouched, preserving their sharpness. Only scanned pages are re-imaged at your chosen DPI for recognition. As long as you don't set a very low DPI (like 150) on scanned pages, quality should remain adequate.
No. The tool only recognizes English, German, and French. You can select one or more of those languages per job, but no other languages are supported. For other languages, you would need a different OCR service.
It's a normal warning that tells you a text layer was added to your file. It also reminds you that OCR is not perfect, so you should spot-check the recognized text, especially on low-quality scans, handwriting, or unusual fonts.
No. The file limit is 500 pages and 50 MB. You'll need to split your PDF into smaller parts and run them separately. Each job accepts exactly one PDF file.
If the OCR didn't find enough text (for example, because the scan was very low, the pages were handwritten, or in an unsupported language), the tool might fail instead of returning a faulty file. If you get a result but text isn't selectable, re-run with a higher DPI (300 or 400) and ensure the language is correct.
Your file is uploaded to private object storage and is automatically deleted after 1 hour if you are not signed in, or after 7 days for signed-in accounts. Until then, it's only accessible to you.
No. The tool accepts the whole file as is and processes each page as needed. However, pages that already have a searchable text layer are passed through untouched, so you might not need to split the file if only some pages are scans.