Can I practice reading my own book or PDF instead of a library passage?

Yes — you can upload a PDF or paste in plain text, and sayit turns it into a practiceable reading exercise: split into chapters, broken down into readable paragraphs, and scored the same way any library passage is, word by word. It's the "bring your own material" option next to the curated reading library, for when you'd rather practice on a book or article you actually care about.

30-second version: Upload a PDF or paste text, and sayit extracts and cleans it up first — real PDFs are full of artifacts like hyphenated line breaks, page numbers, and repeated headers, all of which get stripped or reflowed before you ever see the text. It then detects chapter boundaries conservatively, so a noisy document doesn't shatter into hundreds of bogus chapters, and falls back to evenly-sized sections if it can't find real ones. Each chapter gets a word count and a difficulty level, and you read it paragraph by paragraph with full per-word scoring.

What actually happens to a PDF when I upload it?

More cleanup than you'd expect. Raw PDF text extraction is genuinely messy — words split across a line break with a hyphen, lines hard-wrapped in the middle of sentences, page numbers and running headers mixed into the body text, sometimes outright OCR garbage. sayit de-hyphenates split words, collapses irregular whitespace, and reflows hard-wrapped lines back into real paragraphs before any of it becomes a reading exercise — so what you actually read is close to how the text was meant to flow, not a literal dump of the file's internal line breaks.

How does it know where chapters start?

Conservatively, on purpose. It looks for real heading patterns — "Chapter 7," "CHAPTER VII," "Part II," that kind of explicit marker — distinguishing an actual heading from a sentence that merely happens to start with a similar word. It also recognizes standalone Roman numerals, but only when they form a genuine ascending series, the classic convention in public-domain books, rather than treating every stray Roman numeral as a chapter break. If a document doesn't have enough trustworthy boundaries — or looks like it would shatter into heading-noise — sayit falls back to grouping the text into evenly-sized sections instead, so the material stays practiceable either way rather than failing outright.

Does each chapter get a difficulty rating like the library passages?

Yes — the same beginner, pro, legend classification the reading library uses is computed per chapter, from a word count and readability read on the actual text, so you can see at a glance whether a chapter is a light read or a genuinely demanding one before you start.

How do I actually practice once it's imported?

Chapter by chapter, paragraph by paragraph — the same way a library passage works. Each paragraph is scored through the standard reference-mode pipeline: per-word verdicts, target-versus-produced IPA on anything flagged, and the same accuracy and fluency breakdown any other take gets. Nothing about the scoring changes because the text came from your own upload rather than the curated library.

Can I paste text directly instead of uploading a file?

Yes — plain text works the same way as a PDF upload, minus the PDF-specific cleanup step (since there's no page-layout artifacts to remove from text you typed or pasted yourself). Either path ends at the same place: chapters, paragraphs, difficulty ratings, and full scoring.

Importing at a glance

Try it

Open Books in sayit and upload a PDF or paste in an article you actually want to read — it'll come back split into practiceable chapters within moments, ready to read aloud and score.