How to Check Accessibility

What ScholAccess Fixes

When you upload a file, ScholAccess automatically detects and remediates accessibility issues across six categories:

• Images & Alt Text — Every image, diagram, chart, and equation gets an AI-generated description so screen readers can convey visual content to students. Descriptions are written in plain language; math is rendered in spoken form (e.g., "the integral from 0 to pi of sine x dx") for natural listening.
• Contrast — Text that does not meet the WCAG 2.1 AA contrast ratio (4.5:1 for normal text, 3:1 for large text) is detected and recolored automatically. This covers solid-colored text on solid backgrounds; gradient or image backgrounds are flagged for manual review.
• Structure & Titles — Missing document titles and slide titles are generated so screen readers can navigate by heading. Table header rows are marked, default section names are renamed, and decorative shapes are flagged appropriately.
• Formats — PowerPoint (.pptx), Word (.docx), and PDF files are all supported. Scanned and handwritten PDFs receive page-level AI descriptions. Embedded audio and video in PowerPoint get automatically transcribed closed captions (WebVTT).
• Check Mode — Upload a file in "check only" mode to get an accessibility report without modifying anything. The report lists every issue by category and severity, along with a numeric score you can use to track progress over time.
• Preservation — Your original content, formatting, animations, and media are preserved. ScholAccess adds accessibility metadata (alt text, tags, titles, captions) without altering your design or layout.

What to expect

Our goal is to eliminate all machine-checkable errors and reduce warnings to those that genuinely require human judgment. However, keep the following limitations in mind:

• Reading order is always flagged on complex slides as "Needs manual check." This is normal — automated tools cannot determine the author's intended reading sequence for non-linear layouts. Review it visually and adjust if needed.
• Color contrast may be flagged as "Needs manual check" even after remediation. Our tool fixes contrast for solid-colored text on solid backgrounds, but gradient or image backgrounds require manual review.
• Alt text quality is best-effort. AI-generated descriptions should be reviewed for accuracy, especially for technical diagrams, equations, and domain-specific images.
• Detailed image descriptions — for complex images like charts, diagrams, and data visualizations, ScholAccess adds a hidden off-slide text box containing a detailed description for screen readers. This goes beyond the short alt text (which says what the image is) to provide the full information a sighted user would get from the image. These boxes are invisible during presentations but appear in the slide's shape list. They are labeled a11y_image_explanation and should not be deleted. Microsoft's built-in checker does not do this — it is a best practice recommended by WCAG for complex visuals.
• Table structure — complex tables with merged cells or nested headers may need manual adjustment after remediation.

Scanned & handwritten documents

Scanned PDFs and handwritten lecture notes (e.g., tablet-written or photographed notes) are common in learning environments. Here's what to expect after remediation:

• What we do: Each page is analyzed by an AI vision model that generates a text description of the page content, including handwritten text, diagrams, and equations. This description is added as alt text on the page's image, making the content accessible to screen readers.
• Document title is inferred from the filename and the first page content using AI, since scanned documents typically have no embedded title metadata.
• Math and equations in handwritten notes are described in spoken form (e.g., "the integral from 0 to pi of sin x dx") rather than LaTeX or MathML, which is more natural for screen reader users.
• Remaining checker warnings: Adobe Acrobat may still flag "Image-only PDF" since the document contains no embedded text layer — this is expected for scanned content. The accessibility checker may also report "content not associated with structure" on some pages; screen readers will still read the alt text correctly.
• Accuracy: AI-generated descriptions of handwriting are best-effort. Review them for correctness, especially for technical notation, variable names, and domain-specific terminology.
• For best results, consider using your device's built-in accessibility features (e.g., Live Text on iOS/macOS, Google Lens) alongside the remediated PDF for real-time handwriting recognition.

Verifying Accessibility in Your App

After remediating your files with ScholAccess, use the built-in accessibility checker in each application to review any remaining issues: