Why doesn't Assist's read-aloud function read Aria descriptions?

1. Difference to screen reader functionality

Eye-Able Assistant does not work like a traditional screen reader.

It is designed for users who can still perceive websites visually and supports them by primarily reading aloud visible text.

Text stored in the background – such as in aria-description or aria-describedby – is therefore not taken into account, as it is not relevant to this target group.

2. Why aria-hidden works, but other labels do not

The attribute aria-hidden is taken into account by the assistant, as its purpose is to hide content from users.

One exception is aria-label:

this attribute is taken into account because it is often the only text associated with an element (e.g. in icon buttons).

Other attributes such as aria-description are currently not evaluated.

3. Difference between tab focus and mouse pointer read-aloud function

• Tab focus: Here, the assistant reads visible text elements directly.
• Mouse pointer read-aloud function: This variant is based on the actual recognisable content under the cursor. If there is only an icon without visible text, the function remains silent.

4. Making content specifically accessible for the assistant

In order to have content read aloud even without visible text, it is recommended to use an aria-label.

This is the only attribute that allows the assistant to take invisible text into account.