screen reader
A screen reader is software that says a document aloud for somebody who cannot see it.
In more detail
Whoever cannot see the screen hears it instead. The software moves through the document the way somebody sighted moves their eyes, and says each part in turn: this is the title, this is a list of three things, this is a table of four columns.
For that to work the file has to say what its parts are. A page drawn as marks and nothing else is, to such software, a heap of letters in whatever order the machine happened to put them down: two columns read straight across, a page number in the middle of a sentence, a picture passed over in silence.
What it does on a real document
A payslip goes out to eight hundred members of staff, one of whom is blind. It is one file and it has to serve both: the same page on paper, and inside it the hidden copy of itself saying the order to read it in and what stands in for the picture at the top.
This is also what public bodies across Europe are required to supply, and what a document claiming the accessibility standard is measured against. A document that reads aloud correctly is the whole point of that standard.
Where it comes from, and why
The first programs that said a screen aloud date from the 1980s and worked on plain text. Two columns, or a table, defeated them outright: what they read out was the order in which the letters had been sent to the screen.
The answer, inside a PDF, was to let the file carry a second copy of itself made of names rather than marks. That arrived with the format in 2001, and the accessibility standard published in 2012 turned it from something a document might have into something it must have.
See it on a finished document
A document that can be read aloud A list whose pictures speak too