How to Make LaTeX Documents Accessible: A Guide for Professors

If you teach or publish at a university, you've probably heard the push: all course materials need to be accessible. Slides, handouts, exams — anything a student might need. The problem is that most academic content lives in PDFs produced by LaTeX, and those PDFs are essentially invisible to screen readers. Equations are meaningless glyph sequences. Figures have no descriptions. There's no document structure at all.

This isn't a niche concern. Accessibility offices at universities across the US, UK, EU, and Australia are now requiring compliance with WCAG 2.1 and PDF/UA standards for all published materials. Professors who've been writing LaTeX for decades are suddenly being told their PDFs don't meet requirements — and the tooling to fix it has historically been painful.

That's starting to change.

The Core Problem: Handwritten Notes and Scanned Materials

The most inaccessible academic materials aren't typed documents — they're handwritten lecture notes, whiteboard photos, scanned problem sets, and annotated textbook pages. These are images. A screen reader sees nothing. A student with low vision gets a blank page.

Hundreds of professors at top universities now use Underleaf to solve this. The workflow is straightforward:

1. Photograph or scan handwritten notes — lecture notes, problem sets, derivations on a whiteboard, annotated PDF pages
2. Upload to Underleaf's Image to LaTeX — the AI extracts all text, equations, tables, and diagrams into clean, editable LaTeX source code
3. Edit and compile — fix any errors, add alt text for figures, compile into a clean, structured PDF that screen readers can actually parse

For existing PDF documents (scanned textbook chapters, old problem sets, legacy course materials), the PDF to LaTeX tool does the same thing — extracts the content back into editable LaTeX, preserving equations, tables, and document structure.

The result is that a professor can take a stack of handwritten lecture notes that were completely inaccessible and, within minutes, have typed LaTeX source that can be compiled into a structured, searchable PDF — or converted to HTML for full screen reader support.

Why Standard LaTeX PDFs Fail Accessibility Checks

Even when you type LaTeX from scratch, the default PDF output has accessibility problems:

• No tag structure — Screen readers can't tell headings from body text from captions. The PDF is a flat stream of positioned characters.
• Math is opaque — Equations are rendered as individual glyphs. A screen reader sees the characters "x", "2", "+", "y", "2" — not "x squared plus y squared."
• No alt text on figures — Images and diagrams have no text descriptions.
• Broken copy-paste — Many LaTeX fonts don't map glyphs to Unicode, so copying text from the PDF produces garbled output.

These aren't edge cases. They affect every equation, every figure, and every table in a standard LaTeX document.

Making Your LaTeX PDFs Accessible

There are practical steps you can take today, ordered from quickest to most thorough:

1. Add Document Metadata

This takes 30 seconds and immediately improves how assistive technology handles your document:

2. Add Alt Text to Equations (axessibility Package)

The axessibility package automatically embeds the LaTeX source of each equation as alt text in the PDF. It's one line in your preamble:

Now every equation has readable alt text. It's not perfect — screen readers will read LaTeX source code rather than natural language — but it's a massive improvement over nothing.

3. Add Alt Text to Figures

Every \includegraphics should have a description. With the tagpdf package:

Write alt text that describes the finding, not just the chart type. "Loss decreasing from 2.4 to 0.3 over 100 epochs" is useful. "A line chart" is not.

4. Full Tagged PDF with tagpdf

For complete PDF/UA compliance, the tagpdf package (by the LaTeX Project) creates properly tagged PDFs where headings, paragraphs, lists, and tables are all machine-readable:

Compile with lualatex for full tagging support. This is the most thorough option but requires testing — not all packages are fully compatible yet.

The Workflow That Actually Works

Here's what we see professors doing in practice:

• Before each lecture: Upload handwritten notes to Underleaf's Image to LaTeX. Edit the output. Compile with axessibility and hyperref. Post the accessible PDF to the course LMS.
• For legacy materials: Run old scanned PDFs through PDF to LaTeX. Clean up the output. Recompile with accessibility packages.
• For exams and problem sets: Write in LaTeX as normal, add axessibility to the preamble, add alt text to any figures. Five minutes of extra work per document.

The key insight is that the hardest part isn't the accessibility packages — it's getting handwritten or scanned content into editable LaTeX in the first place. That's the bottleneck Underleaf removes.

Using Custom .sty Files for Consistency

If you're making multiple documents accessible, create a custom .sty file that bundles your accessibility setup:

Then every document just needs \usepackage{accessible-setup} and you get consistent accessibility across all your materials. Underleaf's Image to LaTeX and PDF to LaTeX tools also accept custom .sty files, so the AI output will use your macros automatically.

Going Further: HTML Output with MathML

For maximum accessibility, the best format isn't PDF at all — it's HTML. When you convert LaTeX to HTML, math can be rendered with MathJax or KaTeX, which automatically generates MathML and ARIA labels. A screen reader can then say "x squared plus y squared equals z squared" in natural language instead of reading glyph codes.

Tools for converting LaTeX to HTML with MathML:

MathML is the W3C standard for representing math on the web. All modern browsers render it natively, and screen readers like JAWS, NVDA, and VoiceOver support it. If your institution publishes course materials online (Canvas, Moodle, a course website), HTML with MathML is the gold standard for accessible math.

The practical approach many professors take: distribute the PDF for print and general use, and provide an HTML version for students who need screen reader access. Both can be generated from the same LaTeX source.

What's Coming Next

The LaTeX Project is actively building accessibility into the LaTeX kernel itself. The tagpdf package is the foundation, and the goal is that future versions of LaTeX will produce tagged, accessible PDFs by default — no extra packages needed. We're not there yet, but the trajectory is clear. In the meantime, the combination of AI-powered transcription (getting content into LaTeX) and accessibility packages (getting accessible output from LaTeX) is the most practical path forward.

For a deeper dive into the technical details, see our complete guide to LaTeX accessibility in the Learn section.