You can do the same at home
, just grab the text version of the scan:
and use it with a TTS program.
It is the OCRed text that serves as a basis for the audio and search feature (for example, search for OQEl that you can find in the text version but does not make sense in English).
Still, I find it technically impressive how they manage to link the scanned page (images) with the OCRed text.