Accessible PDF Files

PDFs are one of the most common file types instructors upload; and also one of the most inaccessible. Scanned readings, untagged exports from Word or Google Docs, and older course files often receive the lowest Ally scores in bCourses because assistive technologies cannot recognize structure, text, or images. Many PDFs are created by scanning hard-copy documents, resulting in image-based PDFs. These image-based PDFs are completely inaccessible without converting them to text using Optical Character Recognition (OCR) software. That being said, when possible, consider an alternative format: 

• bCourse pages (HTML): Use bCourses Pages (HTML). For syllabus information or assignment instructions, copy the text directly into abCourses Page. To make sure that your bCourses page is accessible, visit our Accessible bCourses Sites page. 
• The original document:If you created the material, the most accessible version is usually youroriginal file, not the exported PDF. To ensure your original documents are accessible, see the RTL resources for Accessible Documents (Microsoft Word & Google Docs) and Accessible Slides (Microsoft PowerPoint & Google Slides). 
• Library-hosted materials: If you find yourself faced with the need to scan readings, check the Library or publisher website. Many journal articles and book chapters are already available online in accessible formats. Linking to the official publisher resource usually avoids the need for scanning.

Note: If you must scan a document, avoid creating image-only PDFs. When text is locked inside an image, screen readers cannot access it. Use Library scanners or other scanners that offer OCR (Optical Character Recognition), which converts image-based text into readable, searchable text and dramatically improves accessibility.

It’s essential to note that we can’t determine accessibility simply by visually examining a PDF. PDFs must be created using an inbuilt structure called "tagging" so screen readers can navigate them properly. For instance, visually, a text might look like a heading because it’s just larger or bolded, but if it isn’t properly tagged as a heading, a screen reader won't recognize it as such. Untagged PDFs pose significant accessibility issues, but scanned PDFs are even worse, as scanned pages are recognized as images rather than text, making them completely inaccessible.

• Selecting Text: Try selecting text within the PDF. If you can’t select individual text and instead select the entire page, it's likely an image-based PDF, making it inaccessible.
• Looking at your Ally indicators in bCourses: Use the Course Accessibility Report in bCourses to find the worst offenders. Click the Content tab and look for PDFs with Red or Orange scores. Clicking the Ally meter next to one of these low-scoring PDFs helps clarify the specific issue Ally has detected. Common messages include:

  • “This PDF is scanned”, indicating the document is image-based and the text cannot be read or navigated by screen readers.

  • “This PDF is untagged”, meaning the document lacks the semantic tag structure (tag tree) that allows users of assistive technologies to navigate headings, lists, tables, and reading order.

Clicking “What this means” provides more information about the specific accessibility concern and explains how it affects users of assistive technologies.

• Checking for Tags: In Adobe Acrobat Pro, go to File Properties (Ctrl+D in Windows, Command+D in Mac). In the bottom left corner of the Document Properties dialog, see the “Tagged PDF” field. If you see "No" or "No Tags available," your document lacks the necessary structure for accessibility. Adobe Acrobat Pro also has a built-in accessibility checker. To run it:
  1. open the PDF inAdobe Acrobat Pro.
  2. go toAll Tools > Prepare for Accessibility  > Check for Accessibility.
  3. select the desired options and select,Start Checking.
  4. review theAccessibility Report.

The report lists issues into categories such as Alternate Text, Tables, Lists, Headings, etc.

The most effective way to ensure an accessible PDF is to start with an accessible source document.Accessibility features like headings, alt text, and reading order must be built into the original file (Word, Google Docs, PowerPoint, Google Slides)beforeyou convert it. If you skip this step, the resulting PDF will likely lack the necessary tags to be readable by assistive technology. This is critical because the accessibility of a PDF depends heavily on how the original file is structured. To learn how to prepare accessible source files, see our guides:

• Accessible Documents (Word & Google Docs)

• Accessible Slides (PowerPoint & Google Slides)

In a nutshell, your source file should include:

• Proper heading structureusing built-in styles (Heading 1, Heading 2, etc.).

• Meaningful alternative textfor all images, charts, and graphics.

• Clear, descriptive hyperlinks(not “click here”).

• Legible fonts, good color contrast, and clean formattingwith no text boxes or floating elements.

• Simple tableswith headers (avoid merged cells; no nested tables).

• Built-in accessibility checksrun before exporting (Word, PowerPoint, Grackle for Google Docs/Slides).

Export correctly:

Once your source file meets the accessibility basics, export your PDF using the correct settings:

• Microsoft Word/PowerPoint:Go toFile→Save As→PDF. Ensure"Document structure tags for accessibility"is checked in the options.

• Google Docs: Note that while Google Docs exports retain much of the document’s structure, they may lack full accessibility tagging. For best results, consider exporting to Word first and using Microsoft Word’s accessibility checker before converting to PDF.
• Google Slides: Unfortunately, Google Slides does not yet export fully tagged PDFs. Consider downloading your Slides as PowerPoint (.pptx), open in Microsoft PowerPoint, run the accessibility checker, and export the PDF from there with structure tags enabled.
• Adobe InDesign: Go to File→Export and select Adobe PDF (Interactive) or Adobe PDF (Print), then check Create Tagged PDF.
• Never "Print to PDF": Selecting "Print" flattens the document into an image, destroying all accessibility tags. Always use Save or Export.

Due to the complexity of PDF accessibility, it’s often simpler to provide the content in HTML or accessible Word documents whenever possible. If you lack the original source file or a more accessible alternative, remediating an existing PDF may be your only option.

• For untagged PDFs or image-based scans: We recommend starting with Step 1: Automated Prep with Sensus Access. While tools like Ally or Adobe Acrobat Pro are available, SensusAccess generally yields the best results with the least manual effort for these specific issues.

• For PDFs that are already OCRed and tagged: You may skip directly to Step 2: Manual Review with Adobe Acrobat Pro.

Step 1: Automated Prep with Sensus Access

Use this step for image-based scans or untagged documents.

1. Go to UC Berkeley Sensus Access Portal 

2. Click Choose File (find your PDF on your computer)

3. Click Upload

4. Select Accessibility Conversion and select PDF – tagged PDF (image over text). This keeps the visual look of the original page (images/layout) while placing a layer of digital text underneath for screen readers. 

5. Enter your Email address

6. You should receive an email with your new, tagged PDF within several minutes. 

Watch the SensusAccess tutorial,(link is external) which demonstrates how to convert course materials into accessible formats. 

Note: Depending on the quality of the source file, conversion results may vary greatly. Converted files often still need manual fixes (headings, alt text, etc.)  To achieve the best conversion results, read the practices outlined in Conversion Best Practices.

If you have any questions about SensusAccess, please feel free to email sensusaccess@berkeley.edu(link sends e-mail)

Step 2: Manual Review with Adobe Acrobat Pro 

Use this step after SensusAccess OR for PDFs that are already tagged but need alt text or additional remediation.

1. Open your PDF using Adobe Acrobat Pro . (If you need help installing it, see BerkeleyIT instructions for Installing Adobe Creative Cloud

2. Select “Prepare for Accessibility" under the Tools menu (you may need to click View More). 

3. Click Check for Accessibility (and Start Checking) to review the document structure. The panel on the right will detail your accessibility issues. 

4. Go to File > Document Properties allows you to add the Document Title (page 1) and the Language (Advanced tab). These are required for accessibility conformance.

5. Click Add alternative text to add and or review descriptions for images. 

1. Pro tip: Using AI Tools, such as Google Gemini, can help speed up the process of generating initial descriptions. 

2. STEM: Many PDFs include mathematical expressions saved as images, which creates accessibility barriers. See "STEM considerations for math in PDFs" section on this page to learn more.

6. Run Accessibility Checker to confirm that there are no other outstanding accessibility issues. 

7. When you are done fixing the issues, save the file and upload it to bCourses. Your Ally Score for this document should now be green!

Resources to go further: 

• Create and verify PDF accessibility (Acrobat Pro.) 
• Creating Accessible PDFs (LinkedIn Learning). This training is 4.5 hours in length and may be suitable for individuals who frequently need to create accessible PDFs in their role. (LinkedIn Learning is available for free for the UC Berkeley community.)

Mathematical content introduces unique accessibility challenges, especially when materials are exported or scanned into PDF format. Equations, symbols, and notations often end up as images, which screen readers cannot interpret. Even tagged PDFs struggle with complex math. The workflow you choose depends on whether you can edit the original file or if you're remediating a PDF. 

If You Have Access to the Original Source (Preferred Method)

Whenever possible, mathematical expressions should be authored in a format that supports real, accessible math, not images. 

• Option A: Canvas Rich Content Editor (RCE): Canvas supports LaTeX and MathML, which are more accessible than images of equations. Publish equations directly in Canvas rather than embedding them in PDFs. The Math Editor is available directly inside the RCE. To open the Math Editor from the RCE menubar, click Insert, then select the Equation option. You can use the Math Editor for anything from simple equations to more complex notation in upper-division or graduate-level classes. Both instructors and students can access this editor when creating content. For more details, see the Canvas Guides on using the Math Editor(link is external).
• Option B: Produce HTML Files from LaTeX: For documents written in LaTeX, generating HTML output is often easier to remediate for accessibility than PDF output. arXiv (maintained by Cornell University) has adopted this approach and provides guidance on accessible LaTeX workflows using LaTeXML: LaTeX Markup Best Practices for Successful HTML Papers
  
• Option C: Produce Accessible PDFs from LaTeX: It is also possible to produce accessible, tagged PDF output from LaTeX. The LaTeX Project (Using LaTeX to produce accessible PDF) provide documentation on current best practices and ongoing improvements in this area.

At this time, these accessibility improvements do not consistently extend to Beamer presentations, and there is no clear consensus or well-documented workflow for producing fully accessible Beamer output. (If you are aware of effective approaches, please let us know.)

If You Are Remediating a PDF (No Access to the Original File)

If your PDF contains content saved as images, screen readers cannot access the underlying information. Math saved as an image is not accessible. This happens frequently with scanned worksheets, handwritten notes, whiteboard screenshots, or even some exported PDFs. Because these images have no semantic information, assistive technologies cannot read, navigate, or interpret the math correctly. In this case:

• Use AI tools to extract LaTeX or MathML: AI tools such as Google Gemini can transcribe equations from images, including handwritten notes. Upload a screenshot of the equation and use a prompt such as: “Please transcribe the expressions in the attached image using LaTeX.”
• Add the extracted LaTeX to the image’s alt-text: For remediated PDFs, include the LaTeX transcription in the alt text or in a nearby text description. This provides a minimally accessible representation when the original source cannot be updated.
• AI for Complex Graphs & Figures: AI can help draft alt-text for complex graphs and figures that are challenging to summarize, providing a useful starting point for review and refinement.  

Step-by-Step Workflow for using Google Gemini to generate alt-text:

1. Open Gemini: Sign in to your bConnected suite. Select the Google Apps grid (upper-right) and launch Gemini. (Important: To protect sensitive data, always sign in with your UC Berkeley (CalNet) account to use campus-licensed AI tools. This ensures your content is not used to train external models.)

   

2. Capture the Image: In your document (Acrobat, Word, or PowerPoint), use a screen-capture tool (Mac: Shift-Command-4; Windows: Win+Shift+S) to copy the image to your clipboard. 
3. Generate Alt Text: Paste the image into the Gemini prompt area and type: "Provide concise alt text for this image". Press Enter.
4. Review and Transfer: Verify the output for accuracy and instructional relevance. Click the Copy response icon and paste the text into the alt text field of your application.

Advanced Use Cases

• STEM Equations: Upload a screenshot of a mathematical expression and prompt: "Please transcribe the expressions in the attached image using LaTeX."

• Batch Processing: If working with multiple images, you can ask Gemini to generate descriptions for all of them at once.

Many instructors share digital scans of handwritten notes. Unfortunately, these are often "image-only" PDFs that are invisible to screen readers. We suggest converting these notes into an accessible format using AI tools like Gemini or NotebookLM. 

Recommended Workflow:

1. Upload: Upload your scan (PDF or Image) to Gemini.

2. Prompt: Use the following prompt to generate accessible code:"Please generate Canvas-friendly HTML. Use only clean <p>, <h2>, <table>, <ul>, and <code> tags. Do not use external CSS, inline styles, spans, or divs. If there is mathematical notation, convert it to MathML. Structure the content clearly for a Canvas Page".

3. Transfer: Copy the output and paste it into the HTML Editor ( View > HTML Editor)
    

4. Review: Always verify the converted text against the original. AI can misinterpret symbols in messy handwriting.

Note on Visuals: This method effectively converts text and math, but it cannot yet accurately recreate figures, diagrams, or complex graphs. These elements should still be provided with high-quality Alt Text or described textually.)