Describing your images: alt text, and why it can also be heard

The "alt text" field of an image is the most often empty spot on a site, and one of the most regulated. WCAG 1.1.1, echoed by the RGAA and by the European Accessibility Act, asks for something simple to state and tedious to keep: every meaningful image must have a text alternative. In practice, photos still get by, but infographics, charts and maps almost always arrive with nothing useful, because writing a good description by hand, image by image, takes time nobody has. The field stays empty, or worse, is marked "decorative" when it is not.

A short alt and a long description are not the same thing

The first confusion to clear up is that there is not one alternative but two distinct needs. The alt attribute, read by a screen reader in place of the image, must be short: a sentence that says what the image shows and why it is there. "Chart: sales double between January and March" is enough, and an alt that runs long tires more than it helps. But a rich infographic holds information that this sentence cannot carry: the figures, the steps, the legend. For that content you need a detailed description, longer, available next to the image.

That distinction is what WeDispatch produces in one pass: from the image and an optional context (the article's topic), it returns a short alternative ready for the alt attribute, and a detailed description for complex visuals. The context is not a detail: the more precisely you say what the article is about, the more the description sticks to the point rather than to the mere look of the image.

What it does not do, and why saying so matters

A generated description is a solid first draft, not a final text. On a neutral photo it is often publishable as is; on a sensitive or ambiguous visual it needs a review, and you keep control over the final wording. A tool that claimed to fully replace the writer here would be over-promising, and accessibility is precisely a field where false compliance is worse than an honest absence: an automatic, wrong alt ticks the box while misleading the very person who depends on it. The right use is to give the newsroom back the time to review rather than the time to write from scratch. That honesty about the limits is the same line as the French text to speech page: say what the tool does, and what it does not do yet.

The image you can hear

There is one last step, often forgotten. A detailed description serves the person using a screen reader; but the listener playing your whole article reaches the image and hears nothing, because the image is not in the audio stream. WeDispatch can also generate the description as audio, with the same voices as the rest of the article. An infographic then becomes genuinely listenable, consistent with the page's sound, instead of a silent gap in the middle of an otherwise careful reading.

It is the natural extension of audio as an accessibility gesture, which we approached from the regulatory angle in the audio version and screen readers, and from the use angle for low-vision and partially sighted readers. Describe the image, then make it audible, and you close a loop that text alone left open.

Where to start

No need to redo everything at once. Spot the pages where visuals carry the information (data stories, illustrated features, product sheets with a diagram) and handle those first: that is where the text alternative is most missing, and where it pays off most. The gesture is three steps: you drop the image and the context, you get the short alt and the long description, and you voice the description if your edition already runs audio. It fits the same effort as subtitles drawn from the audio: start from the content you already have, and make it accessible through several doors, so nobody is left in front of a silent image.

Read next

• Reading times and opening hours aloud in French
• Non-English foreign words in French text read aloud
• Parentheses and asides: what French text to speech does with them