By turning on the "Manually crop pages before extracting text" option in @Voice PDF import screen, and pressing the "Open file" button, the PDF Plugin will start and show the first page of your PDF document. You may change pages with the slider and < > buttons on top. For each page you may move the gray area edges up and down, right and left as needed, to cut out any unwanted text, such as page numbers, headers and footers, sidebars, footnotes etc. Only the text from the white area in the middle will be included when extracting text for reading aloud.
The additional functions on the "hamburger" menu let you propagate current page crop to other pages, exclude some pages completely from the text extraction process etc. When finished, press the Finish item there. This will take you back to the main @Voice app, start the actual text extraction, followed by reading aloud.