Drop in an EPUB and check it against the spec first — right in your browser, without uploading it. Where a fix is certain to be right, repair it. Complete its metadata and cover, by hand or from an ISBN. Then modernize it to a standards-compliant EPUB 3.3: structure upgraded, images kept as they are, or re-encoded to WebP if you choose. You can also archive a book to a compact .eparc and restore it later, or import a PDF into a reflowable EPUB. No install, no account.
Target version
Conversion options
Experimental. Shrink a book into a compact .eparc archive — the text is losslessly compressed and the images are kept untouched. Your book is fully recoverable: switch to Restore anytime to get a working EPUB back.
An .eparc is an archive, not a readable e-book — it won't open in a reader. Keep it for storage and Restore it when you want to read.
Turn a PDF into a reflowable EPUB. Works for PDFs that have a text layer — born-digital books and scans that are already searchable. Scan-only PDFs need OCR, which is coming in a later release.
Experimental. Quality depends on the PDF; tables and equations are best kept as images for now.
Let an AI OCR service read a PDF — including scans and photos — into clean text, then build a reflowable EPUB. Your file is sent to the provider you choose; this is the only mode that leaves your machine. It uses the server's own API key.
Experimental. Your PDF is sent to the selected AI provider for OCR; epublift stores nothing.
Experimental. Turn an .eparc archive back into a working .epub. By default you get your book back exactly as it was archived.
Size-safe: an image is re-encoded only when WebP is actually smaller — your book never grows. Low-quality images aren't upscaled.
Image format
_v3.3.kepub.epub. Uses the image format chosen above — for WebP on Kobo you need the WebP plugin; without it, pick Keep original.
Fix a book's metadata — edit by hand, or auto-fill the gaps from Open Library by ISBN. Filled only in the book's own language. Drop an .epub above to begin.
Check an .epub against the spec — with epubcheck-compatible message IDs (RSC-005, OPF-…). Powered by epubveri, a pure-Rust, JVM-free validator.
🔒 Validated entirely in your browser with WebAssembly — your file is never uploaded. A quick pre-check, not an official conformance certificate.
Beta · epubveri is pre-1.0 and follows the EPUB specification, which is not always what epubcheck reports. If a retailer asks for epubcheck, cross-check with epubcheck.
Fix what Validate finds, wherever a fix is certain to be right: undeclared entities, invalid ids, broken spine and manifest entries, empty metadata and more. Powered by epubsana. Your original file is never modified.
🔒 Repaired entirely in your browser with WebAssembly — your file is never uploaded.
Safe fixes are selected for you. A fix that needs a decision shows why — tick it if you agree. A fix that makes the book worse is undone automatically.
version="3.0"dcterms:modified timestampnav.xhtml from toc.ncxYour file is converted in memory and deleted immediately. We never store, log, or read your books.
Prefer to self-host? docker compose up and you have your own private instance.
Built in pure Rust, licensed AGPL-3.0. Read the source →