Small tools for the work between the big things
One useful thing at a time
A sentence trapped in a screenshot. A scanned book page you want to quote. A PDF whose words cannot be selected. FeedPDF starts with these ordinary tasks and tries to make them easier.
A focused first release
Image OCR, scanned PDF OCR, Hindi OCR and Sanskrit OCR share the same simple workspace. You can select PDF page ranges, control column order and download each page as a UTF-8 text file. Translation and audio are outside this release.
Designed to leave less behind
No account is required. Uploads and extracted text live only in temporary memory with a 10-minute limit. The OCR workspace has no advertising or analytics scripts.
Technology and limits
Recognition uses self-hosted PaddleOCR with separate English and Devanagari models. PDFs are rendered locally on the server; documents are not sent to an external AI provider. Printed text works best, and Sanskrit still needs human review.