The media library is where accessibility debt piles up quietly. A site that started with twenty images now has three thousand, most uploaded with names like IMG_4471.jpg and no description. The marketing team posts a video every month, relying on whatever captions the platform generated. Board minutes, menus, price sheets and brochures sit in a folder of PDFs, some of them scanned from paper. None of this shows up as a broken page, yet a screen reader user meets silence or nonsense on every one. Our media accessibility services deal with that backlog in a sensible order: images, video, audio and documents, prioritized by what visitors actually use, and then a routine so new uploads arrive in good shape.
What our media accessibility services cover
Media is anything on the site that is not plain text: photos, logos, icons, charts, infographics, video, podcasts and audio clips, slide decks, and downloadable documents such as PDFs and Word files. Each type has its own WCAG requirements, and each type fails in its own predictable way.
| Media type | What WCAG asks for | What usually goes wrong |
|---|---|---|
| Images | A text alternative that serves the same purpose (1.1.1 Non-text Content) | Missing alt, file names read aloud, decorative images described at length |
| Prerecorded video with sound | Synchronized captions (1.2.2), plus audio description at Level AA | Uncorrected auto captions, no description of on-screen text |
| Audio only | A text alternative such as a transcript | Podcast episodes with no transcript at all |
| Autoplaying or moving media | A way to pause, stop or hide it (2.2.2) | Background videos and animated banners with no pause control |
| Documents | The same principles as pages: structure, alt text, reading order, contrast | Scanned PDFs, untagged files, tables built from tab stops |
We handle all of these, but rarely all at once. Most sites have one category that is clearly the worst, and the audit tells us which.
Alt text at scale, not one image at a time
Adding alt text to a single image is easy; our tutorial on adding alt text in WordPress walks through it. The hard part is a library of thousands with no description, uploaded over years by different people, used in posts, pages, product listings and widgets.
We approach it as a triage job:
- Export and map. We list every image with its current alt text, file name, where it is used and how often those pages are visited.
- Sort by purpose. Each image used on a page is marked informative, functional or decorative. A logo that links to the home page is functional; a photo of a team at work in a blog post is informative; a swirl behind a heading is decorative.
- Fix the templates first. Logos, header images, icons and buttons in the theme appear on every page, so one fix there improves the whole site.
- Work down by traffic. High-traffic pages and key journeys such as services, products and contact are written next.
- Batch the long tail. Older posts are handled in batches, with editorial judgment applied to each image rather than an automatic guess.
The media library field versus the image in the post
One thing confuses many site owners. In WordPress, the alt text you type into the media library is copied into a post when you insert the image. Change it later in the library and older posts usually keep the old version, because the text is stored in the post content. So fixing the library alone does not always fix what visitors hear. We check the rendered pages, not just the library, and update posts where needed. How the alt, caption, title and description boxes differ is explained in our post on the WordPress image fields.
Why we do not rely on AI-generated alt text alone
Automatic description tools can say “a group of people smiling in an office.” What they cannot know is that the photo is there to show your new downtown location, or that the chart’s point is that wait times fell. Alt text depends on why the image is on the page. We may use automated suggestions as a starting draft for a large batch, but a person reviews every description that goes live.
Decorative images, charts and infographics
Not every image needs words. Give a purely decorative image alt=”” and assistive technology passes over it. Describing ornaments and stock textures fills a page with noise. Our article on telling decorative images from informative ones sets out how to decide.
Complex images need the opposite: more than a short alt can hold. A chart of quarterly enrollment, a floor plan or an infographic with twelve facts needs a short alt naming what it is, plus the full information somewhere in text, usually right below the image or in a linked data table. We rebuild infographics as real HTML where the content matters, because text in an image cannot be zoomed, searched or translated.
Captions for website videos
WCAG 1.2.2 requires captions for prerecorded video with sound. Captions carry the dialogue plus important sounds: [phone rings], [laughter], [music stops]. They help deaf and hard of hearing viewers, and also anyone watching on a phone in a quiet office.
Most platforms now generate automatic captions. They are a starting point. Names, technical terms, product names and local place names come out wrong, punctuation disappears, and speaker changes are not marked. We edit them against the audio, fix the timing where lines run too long, and upload the corrected file to the video host.
Where the video lives changes the work:
- YouTube and Vimeo embeds. Captions are managed on the platform, so we need access to your channel or account. We also check the embedded player’s controls are reachable by keyboard.
- Self-hosted video. Captions are added as a WebVTT file attached to the video element, and we test that the browser player shows the captions button.
- Social clips reposted on the site. Burned-in captions are better than none, but they cannot be resized or turned off; a proper caption track is preferred.
Embeds also affect speed. If many videos load full players on page load, our note on lighter YouTube embeds shows a pattern that keeps pages fast; we make sure that pattern stays keyboard-friendly and labeled. Caption detail, including captions versus subtitles, is in the post on closed captions for website videos.
Background and hero videos
A looping video behind a home page headline is decorative most of the time, so it does not need captions if it has no meaningful sound. It does need a visible pause button under 2.2.2 if it plays for more than five seconds, and the headline on top of it must still meet contrast at every frame. Many themes add these videos without either.
Transcripts for audio and video
A transcript is the text version of an audio or video file, readable at the reader’s own pace and searchable. For audio-only content such as a podcast, a recorded sermon or a voicemail greeting sample, a transcript is the main accessible alternative. For video, it is a useful extra that helps deafblind visitors using braille displays and anyone who prefers reading.
A good transcript includes who is speaking, meaningful sounds and, for video, descriptions of important visual content. It sits on the same page as the media or one clear link away, not buried in a download. Our post on transcripts for podcasts and audio covers format and placement.
We often produce the transcript and the corrected caption file from the same edited text, which saves effort and keeps the two consistent.
Audio description, in plain terms
Captions help people who cannot hear. Audio description helps people who cannot see. It is extra narration that explains important visual information not conveyed in the soundtrack: text that appears on screen, a speaker’s name card, a product being demonstrated silently, a chart that appears without comment.
At WCAG Level AA, prerecorded video with meaningful visuals is expected to have audio description. In practice, much of it can be avoided at the script stage. If the presenter says “our new location is at 400 Main Street” while the address appears on screen, nothing more is needed for that moment. We review videos for gaps of that kind, recommend script changes for future videos, and for existing ones suggest the least costly fix: a described version, a text description alongside, or a re-edit. We do not record professional narration ourselves; where that is needed we say so and help you brief someone who does.
PDFs, Word files and slide decks
Documents are the media type sites most often forget. The first question we ask about each one is whether it should be a document at all. A one-page menu, a set of opening hours or an event flyer usually belongs on an ordinary page of the site, which is simpler to keep usable and simpler to update.
For documents that must stay as files, such as forms people print, annual reports or policy documents, we check:
- That the file has real, selectable text rather than a scanned picture of a page.
- That it is tagged, with headings, lists and tables marked up so screen readers can follow the structure.
- That reading order matches the visual order, especially in multi-column layouts.
- That images and charts carry alternative text and decorative elements are marked as artifacts.
- That the document language and title are set.
- That color contrast and link text hold up as they would on a page.
Where possible we fix the source file, such as the Word document or design file, and export again, because patching an exported PDF is slow and does not survive the next edit. The post on accessible PDFs for websites explains tagging and checking in more depth.
How a media accessibility project runs
- Inventory. We list images, videos, audio files and documents, with where each is used and how much traffic it gets.
- Audit a sample. We test representative items of each type as part of a wider WCAG 2.2 audit or on their own.
- Agree priorities. Theme media and key journeys first, then high-traffic content, then the archive.
- Fix in batches. Alt text, caption files, transcripts and document repairs, logged item by item. This sits alongside our wider code and content remediation when the theme also needs work.
- Check rendered pages. We confirm the fix appears where visitors meet it, not only in the library or platform.
- Set up the routine. Editor guidance and checks so new uploads do not rebuild the backlog.
What you get and what we need
You receive a media inventory with the status of each item, the fixes themselves applied on the site and on your video platforms, corrected caption files and transcripts you own, repaired or recreated documents, a log of what changed, and a one-page guide for editors covering alt text, captions and documents.
What you will not receive is a claim that your media, or the site, is now “compliant.” Large archives are usually fixed in stages, so the log shows plainly which items are done, which were retired and which older files are still waiting, along with the order we suggest for them.
We need:
- A WordPress user with rights to edit posts and media, and staging access if you have it.
- Access to your YouTube, Vimeo or other video accounts, or someone who can upload caption files.
- Source files for important documents where they exist.
- Someone who knows the subject matter to answer questions about what key images and charts are meant to show.
- A list of the media your customers use most, if you know it.
How to judge whether the media work is good
Pick a page with several images and turn on a screen reader, or use a browser extension that shows alt text. Does each description say what the image is for, briefly? Are decorative images silent? Play a video with captions on and the sound off: can you follow it, including names and numbers? Open a PDF and try to select the text; then use the reading order panel in your PDF reader, or simply listen to it with a screen reader. Finally, upload a new image as an editor would and see whether the guidance prompts the right habit.
Media mistakes we often correct
- Alt text stuffed with keywords. It reads badly aloud and helps nobody.
- “Image of” at the start of every description. Screen readers already say it is an image.
- Trusting auto captions unedited. Errors in names and figures change meaning.
- Transcripts as downloadable files only. Put the text on the page.
- Scanned documents uploaded as the final copy. They are pictures of text, and screen readers have nothing to read.
- Fixing the library and assuming posts update. Check the rendered page.
Doing media work in-house or handing it over
Everyday video captions and alt text are within reach of a careful editor: describing new uploads, correcting captions on a handful of videos, publishing a transcript with each new episode. If your library is small, start there and set the habit.
Outside media accessibility services make sense for a large backlog, for documents that need tagging, for video libraries across several platforms, or when you need a dated record of what was fixed. Ask any provider: will a person review every alt text? Do you edit captions against the audio? Can you fix documents at the source? Do you check the rendered page after each fix? Will we own the caption and transcript files?
Keeping media accessible as the site grows
Every new post, product, video and upload either adds to the backlog or does not. The routine we set up is simple: alt text written at upload, captions corrected before a video is published, documents checked or turned into pages before they go live, and a monthly look at what was added. Our ongoing accessibility checks include new media in that review. For the wider picture of how media work connects to audits, fixes and monitoring, see the main accessibility services page.