What this checker tests, and what it cannot
An accessible PDF is one that a screen reader, a keyboard or a magnifier can use as well as a sighted mouse user can. Some of that can be tested by software, because it is written into the file: whether the document is tagged, whether it has a title and a language, whether images carry alternate text, whether form fields have descriptions. Some of it cannot, because it depends on meaning: whether the reading order makes sense, whether the alt text actually describes the chart, whether grey text on a white background is readable.
This tool runs the first kind of check. It reads your PDF, tests 14 things, and marks each one Pass, Fail, Warning or Not applicable, with a plain explanation and a way to fix it. It does not change your file, and it does not pretend to certify it. A clean report is a good start, not proof of WCAG, Section 508, ADA or PDF/UA conformance. The report ends with the checks you still need to do by hand, and the PDF accessibility checklist for ADA and WCAG walks through them in more detail.
The 14 automated checks
- Security settings. Encryption that blocks access for assistive technology, or an open password, stops a screen reader before it starts.
- Tagged PDF. Tags are the hidden structure (headings, paragraphs, lists, tables, figures) that screen readers navigate by. The check requires both the tagged flag and a structure tree that actually contains tags.
- Real text, not scanned images. Every page is tested for a text layer. Scanned pages are listed by number, because a screen reader reads nothing on them until OCR PDF adds text.
- Document title. A title in the metadata, and the viewer set to show it instead of the file name.
- Document language. The language code that tells a screen reader which pronunciation rules to use.
- Images have alternate text. Every tagged figure is checked for alt text, and obvious placeholders such as "image" or a file name are flagged.
- Headings, tables and links. Heading tags and their levels, header cells in tables, and whether each link is connected to a Link tag and has a description.
- Tab order, form fields and fonts. Pages with links or fields should follow the document structure for tab order; each form field needs a tooltip; fonts should be embedded and map to real characters.
- Bookmarks. Documents over 20 pages should have bookmarks so readers can move between sections.
Why these matter to a real reader is explained in what makes a PDF accessible to screen readers.
Fixing what fails
Most failures are easiest to fix in the source document, not the PDF. Use real heading styles, add alt text to pictures, set the title and language, then export again with the tagged-PDF or accessibility option switched on (Word calls it "Document structure tags for accessibility"; LibreOffice has a "Universal Accessibility (PDF/UA)" export option). If you only have the PDF, PDF to Word can give you an editable copy to repair and re-export.
Scanned documents need text before anything else: run OCR PDF, then check again, and see how to make a scanned PDF readable by screen readers for the rest of the job. If the security check fails, Unlock PDF removes restrictions on files you own. Documents that also have to be archived can go through the PDF/A Checker as well.
When you need more than an automated check
For a formal PDF/UA test, use a dedicated tool such as PAC, the free desktop checker from the PDF/UA Foundation, or the Full Check in Adobe Acrobat Pro, and then test the file with a screen reader. NVDA is free on Windows, and VoiceOver is built into Macs and iPhones. Listening to your own document for five minutes finds problems no checker can.
Continue with Google