Block out areas that shouldn't be read
It would be great if we could see and adjust the regions of a PDF that are being detected by the reading software to correct mistakes.
Example one: Information from the header or footer is being narrated and breaking up paragraphs split between two pages. ("Consequences of [INTERNATIONAL JOURNAL OF PLANT SCIENCES] these differences for plant ﬁtness have not yet been investigated.")
Example two: Random text from figures or boxes are being shuffled into adjacent paragraphs as if being read normally from left to right.
If the "read area" is a transparent green polygon overlaying the page, I'd want to shorten it to exclude the header and footer or draw a red rectangle covering a figure or text box letting the software to ignore this area.
Sounds like a programming nightmare, but I can dream, can't I?