Using RegEx to match and remove random text within PDF

Hi there,
I have a PDF which has text boxes indented within the actual text of a book.  When using @Voice (on Android) with this PDF, the result is that the text boxes will be read even though they are already part of the main text and do not need to be read.  The result is that the flow of a sentence is interrupted and doesn't make sense (unless you're watching the highlighted text as it is read out).
Below is an example of the kind of text I am referring to:
 
"In order to correctly use this program it is recommended that you first read the entire manual.  This is particularly a must for new users!  I cannot stress this enough!! If you've read this manual before, you can however, use it as a reference<NO PERIOD HERE>
<CR>
This is particularly a must for new users!  I cannot stress this enough!!   <<< Repeated text I'd like not to be read
<CR>
and skip to any section you like."  <<< continued text before it was interrupted.
 
In the above example please ignore the comments within <> since they are only to highlight carriage returns <CR> and that a period does not end the original sentence before the <CR>.
 
Is this possible to use RegEx (within Settings/Edit speech) since the software seems to read one sentence at a time and even considers the sentence ended though there isn't a period at the end of the first sentence (as in the example above), but rather a carriage return/newline?
 
Many thanks for your help!
 
Kind regards,
Morden
 

Is there a way we can remove page numbers distributed all across page? They have a fixed format
For ex: 
-1-
-2-
-3-
 
I tried hard coding regex replace, it worked well. Is there anyway to make it dynamic? Something like "-TEXT-" will work I think.
 
Thanks