Hi Katie,
I'm Rodrigo and I will help you.
This is more complex than it looks. Scanned PDFs are like a "picture" of the document and "normal" PDFs are like a Word documents, that can't be edited.
Since it is like a picture, the "read aloud" does not recognize the characters in the picture.
A good option is to convert these files to a "readable" one there are some free services that can do that, they are not perfect, but can help.
https://ocr.space/
https://www.onlineocr.net/
If you want to know more about it, you can search for OCR on the search engine.
You can leave a suggestion to implement this feature in a future update of Microsoft Edge too, Open the feedBack Hub (Shortcut WinKey + F) and send your suggestion to implement OCR to the Read Aloud tool for PDFs.