How do I convert a PDF to an accessible EPUB for screen readers?

Go to toolkit.bot/pdf2epub, upload your PDF, and download the EPUB 3. The output includes semantic HTML5 headings (H1-H3), a navigation document, and logical reading order — the structure screen readers require to navigate properly. For full EPUB Accessibility 1.1 compliance (required by the European Accessibility Act 2026), additional manual steps like adding figure alt text may be needed.

Is EPUB better than PDF for screen reader users?

Yes, significantly. EPUB 3 has a defined semantic reading order, HTML5 heading structure for navigation by heading level, and is built on WCAG 2.1 accessibility standards. PDFs — especially scanned documents and multi-column papers — often have incorrect reading order, no heading structure, and missing alt text, making them difficult for JAWS, NVDA, and VoiceOver users.

Does toolkit.bot produce WCAG-compliant EPUB output?

toolkit.bot produces EPUB 3 with the semantic structure required for WCAG 2.1: HTML5 headings, nav landmarks, and logical reading order. For full EPUB Accessibility 1.1 compliance required by the European Accessibility Act 2026, additional manual work may be needed — adding alt text for figures, auditing complex tables, and running EPUBCheck with accessibility rules enabled.

What screen readers work with EPUB 3 files?

All major screen readers support EPUB 3: JAWS and NVDA on Windows (best via Thorium Reader, which is screen-reader-optimized by EDRLab), VoiceOver on Mac and iOS (via Apple Books), and TalkBack on Android (via Moon+ Reader or Google Play Books). Thorium Reader on Windows has the most complete EPUB 3 accessibility support including EPUB Media Overlays for synchronized audio narration.

PDF to Accessible EPUB for Screen Readers: A Practical Guide for Libraries and Accessibility Teams

Fixed-layout PDF is the single largest barrier to accessible digital content delivery. A tagged PDF can technically pass an automated accessibility checker, but it still fails readers who depend on screen readers, use e-ink devices, or need to adjust font size, spacing, and color. EPUB3 with proper accessibility metadata is a different category of document — one that is genuinely readable by every major assistive technology.

This guide is written for library staff, accessibility coordinators, and content managers who need to produce accessible EPUB from existing PDF sources.

Why EPUB3 Outperforms PDF for Accessibility

WCAG 2.2 Success Criteria and EPUB Accessibility 1.1 converge on the same requirements:

Text must reflow to 400% zoom without loss of content (WCAG 1.4.10 Reflow): PDF fails this for most multi-column and complex layouts; EPUB passes by default.
Reading order must be programmatically determinable (WCAG 1.3.2): PDF reading order depends on tag structure and is frequently wrong; EPUB reading order is explicit in the document structure.
Navigation landmarks must exist (WCAG 2.4.1, 2.4.6): EPUB3 requires a table of contents and supports landmark navigation (chapters, sections); PDF requires manually tagged structure that is often missing.
Language must be specified (WCAG 3.1.1, 3.1.2): EPUB3 requires a language declaration; PDF language tags are often missing.

For readers using JAWS, NVDA, VoiceOver, or Dolphin, EPUB3 produced from a well-structured source PDF provides a significantly better experience than the equivalent tagged PDF.

Why PDF Fails Screen Readers Specifically

PDF renders content as positioned objects on a fixed-size page, which creates three persistent problems for assistive technology:

Reading order is unpredictable. Screen readers traverse PDF objects in file order, which often differs from visual order in multi-column layouts, sidebars and footnotes.
Reflow is impossible. WCAG 1.4.10 requires content to be readable at 400% zoom without horizontal scrolling; a page with fixed physical dimensions cannot do this, with or without Acrobat Pro.
Semantic structure is fragile. Heading, list and table tags in PDF are metadata layered over the visual layout; they break easily and are hard to audit at scale.

In EPUB3 the same properties are native: H1–H6 are real HTML, so users navigate by heading with the NVDA H key or the VoiceOver rotor; DOM order is reading order; xml:lang drives correct pronunciation; and EPUB Accessibility 1.1 metadata is surfaced by reading systems such as Thorium Reader, Apple Books and Kobo before the file is opened. On Windows, NVDA and JAWS work with Thorium Reader; on macOS and iOS, VoiceOver with Apple Books; on Android, TalkBack with Voice Dream Reader or Lithium. EPUB3 is also the native format for OverDrive, Libby and most accessible lending platforms, and works on Kindle (via Send to Kindle), Kobo, Apple Books and Google Play Books.

When to Convert PDF to EPUB

This approach is appropriate when:

A patron requests an accessible version of a PDF document already in your collection
You receive a PDF from a publisher or author that needs to be made accessible before distribution
Staff need to convert supplemental reading materials for courses or training
A document needs to be readable on e-ink devices (Kindle, Kobo) or mobile apps that don't render PDF accessibly

How to Convert PDF to Accessible EPUB

Go to toolkit.bot/pdf2epub
Upload the PDF
Download the EPUB3 file (typically ready in 30 seconds)
Open in a screen reader or EPUB validator to verify output quality

The tool produces EPUB3 files with semantic HTML5 structure (headings mapped to H1–H6), language declaration, alt text for images extracted from PDF alt text where present, table structure preserved where tables are detected in the source PDF, and EPUB Accessibility 1.1 metadata.

What to Check After Conversion

Open the EPUB in Thorium Reader (free, Windows/Mac/Linux) with a screen reader running and verify:

Heading navigation works (NVDA H key, VoiceOver rotor)
Reading order is logical through any multi-column section
Tables are announced as tables with their headers, not read as flat text
Images have alt text or are marked decorative

Then run ACE by DAISY for a machine-readable report; for publicly distributed documents, attach it to your accessibility statement.

Common Issues and How to Handle Them

Headings not detected: If the source PDF doesn't use font size or style consistently to indicate headings, the EPUB may produce a flat structure. For high-priority documents, manually add heading structure in an EPUB editor (Sigil is free) after conversion.
Reading order errors in two-column PDFs: Multi-column academic papers sometimes produce incorrect reading order. Check with a screen reader; if the order is wrong, the document may need manual cleanup.
Scanned PDFs: If the PDF was created from a scan with no text layer, the converter uses OCR fallback. OCR accuracy depends on scan quality. For accessibility-critical documents from scanned sources, a human review of the OCR output is recommended before distribution.
Missing alt text on figures: If the source PDF lacks alt text on images, the EPUB will also lack it. For accessibility compliance, add alt text in an EPUB editor after conversion.

Choosing a Converter: Buyer Checklist

For screen-reader workflows, the best converter is the one that leaves the least manual remediation: correct reading order through columns and side notes, preserved heading hierarchy, low character-drop text extraction, usable table output, and automatic output checks before distribution. Ask every vendor:

Do you verify output quality automatically? Most tools convert and stop; ask whether text fidelity and structure are checked after conversion.
How do you handle two-column and academic PDFs? Request sample conversions and test reading order in NVDA browse mode.
What happens with scanned PDFs? Is OCR included, and how does OCR confidence affect the output?
Can we inspect evidence of conversion quality? Machine-generated reports, not “looks good”.
Do you emit EPUB3 accessibility metadata? Structure and metadata should align with EPUB Accessibility 1.1 and WCAG 2.2 AA.

A 15-minute evaluation: pick three hard PDFs (a two-column paper, a scanned document, a table-heavy report); convert each with every candidate; open the outputs in NVDA or VoiceOver and test heading navigation, reading order and table comprehension; run EPUBCheck and ACE by DAISY on each; score on error rate plus remediation time. The lowest remediation burden wins, whatever the UI looks like. toolkit.bot's differentiator is built-in self-verification: every conversion is checked with text-level and render-level validation so structural drift is caught before your QA does.

Resources for Library Accessibility Teams

EPUB Accessibility 1.1 specification — W3C
ACE by DAISY — free EPUB accessibility checker
EPUBCheck — free EPUB validator

Convert a PDF to accessible EPUB3 — browser-based, free, no account required. Produces EPUB3 with EPUB Accessibility 1.1 metadata.

Convert a PDF →