DocForensic

PDF Alt Text Generator

Find every image and figure in a tagged PDF, flag which ones are missing real alt text, and generate a reviewable description for each before anything is written to the document.

This tool scans a tagged PDF for images and figures, tells decorative images (icons, repeated logos/watermarks) apart from real content, and builds a description for each one from real, extractable signals: the image's size and position, whether it looks like a chart/graph or a photo, any text visible in the image itself (via OCR), and the nearest caption or heading on the page.

It does not use an AI vision model, so it never claims to know what a photograph actually depicts - every description is a disclosed template built from those signals, always shown to you for review and edited or rejected before it's written into the PDF's tag structure. Nothing is changed until you approve it.

Simple workflow

How it works

  1. 1

    Add tagged PDFs

    Add one or more already-tagged PDFs to the list.

  2. 2

    Analyze

    Run the analysis to see every detected image with its suggested description and confidence.

  3. 3

    Review & apply

    Approve, edit, or reject each suggestion, then write the approved alt text and download the result plus an audit spreadsheet.

Frequently asked questions

My PDF has no tags at all - can this tool add alt text to it?

No. This tool attaches alt text to an EXISTING tag structure - it can't create one from scratch. If your document isn't tagged yet, run PDF Accessibility Checker first to confirm what's missing, then tag the document before using this tool.

Does this use AI to actually look at my images?

No. Every description is built from extractable signals only - image geometry, a chart-vs-photo color heuristic, on-image OCR text, and nearby caption text - never a vision model claiming to have "seen" the image. That's disclosed in every generated description's confidence rating, and nothing is written until you review and approve it.

What happens to images that already have good alt text?

They're left untouched. This tool only generates suggestions for images that are missing alt text entirely or whose existing alt text looks like a placeholder (a bare filename or a generic label like "image1").

Stay ahead of your document work

New tools and thoughtful product updates, delivered occasionally. No noise, no spam.