Text recognition from images please!

I often come across PDFs where the text isn't selectable, making it impossible for Audemic to recognize it. The same issue arises with pictures of books. It would be fantastic if Audemic could overcome these challenges and still recognize text in such cases.