Audio on a school website: accessibility, and the data question

A primary school, secondary school or college website publishes texts that everyone is supposed to read and that many never finish. The note to families about the school trip, the circular on term dates, the school rules, the education project: they tend to be long, sometimes administrative, and they address a very wide audience. Parents who read with difficulty, families who speak another language at home, pupils who are dyslexic or still learning to read fluently. For each of them, a "listen" button at the top of the page concretely changes access to the information, while taking nothing away from anyone else: the text stays readable for those who prefer to read.

It is the same need we describe in our piece on audio and dyslexia, moved into a setting where the reader did not choose to be there: a family does not subscribe to the school website, they have to visit it because it is the official channel. That makes accessibility less of an option than a duty of service.

The constraint specific to schools: minors among the readers

Where a news site or a company measures its audience without much soul-searching, a school cannot. Among the people who will listen to an article, there are pupils, and therefore minors. This is not a theoretical point: it decides what a tool is allowed to record. Before handing the school website to any speech tool, the real question is not "is the voice pleasant," it is "what does this tool note about the person listening." This is exactly the kind of point we lay out on our French text-to-speech page, and it deserves to be checked line by line rather than taken from a brochure.

What our player records about a listener

Here is what the player notes when someone listens to an article, read from the code that does it, with nothing added. Each recorded event has the shape "this article, this type of event, at this time." The types are few: a listen started, and four milestones that say how far the listen went (a quarter, half, three quarters, the end). Nothing else.

What matters most is what is not there. There is no listener identifier at all. The player does not place a marker on the person listening, does not recognise them from one page to the next, and does not track their playback position. This is a design decision, not a setting someone could forget to switch on: the listening minutes shown to a publisher are an estimate deduced from the milestones, precisely because the real position is never followed. A pupil who listens to the school rules three times is not seen as a pupil who listened three times; those are three listens started, with no face behind them.

There is a second safeguard, and it protects a person rather than a statistic. The breakdown of an audience by origin (how many listens came from a letter, how many from the site) only appears above a certain number of listens. Below that, it stays hidden, because an exact percentage over six listens does not describe an audience: it points to one person. We would rather show nothing than a figure able to single someone out. For a school whose numbers are sometimes small, that threshold is not a detail.

What it does not solve

Audio makes a text accessible to the ear; it does not make the rest of the site compliant. A player that cannot be driven from the keyboard, whose focus is not visible, or that ignores a request to reduce motion, makes accessibility worse rather than better. That is a subject in its own right, which we treat elsewhere, and it holds for a school as for any public site. Honesty also means saying that text-to-speech is not a replacement for the work of a specialist teacher: it is a tool of comfort and access, useful precisely because it is simple and asks nothing of the person listening.

Finally, a school should know where the text it voices goes, just like any organisation. We explain this in detail in the article on where your text goes when you send it to text-to-speech. For a note to families the content is often nothing sensitive; but the discipline of asking the question once, at the right moment, beats never asking it.

In short

For a school website, audio answers a real need for access: families and pupils for whom reading is an effort. What makes it particular is that the audience includes minors, which shifts the question from comfort to data. On that point, the player records listens and aggregate milestones, with no listener identifier and without tracking playback position, and it hides any breakdown fine enough to point at a person. That is what it does today, and it is all it does.

Read next

• Turn your feed into a podcast on Apple, Spotify and Deezer: what each one requires
• French text-to-speech and regional accents: can you read an article with a Marseille accent?
• Writing for the ear: what reads well and listens badly