Generating an audiobook with AI can look deceptively simple:

Extract the text.

Send it to a text-to-speech system.

Export the audio.

That workflow works reasonably well when the source is a clean EPUB, HTML file, or Word document. It becomes much harder when the book exists only as a scanned or layout-heavy PDF.