RSVP Reading for PDFs

RSVP and PDFs are a genuinely awkward combination, and it's worth understanding why before you pick a tool — because most PDF speed readers fail at the import step, not the reading step.

A PDF is not text. It's a page-description format: instructions for placing glyphs at coordinates. Extract the raw text naively and you get the author's words interleaved with everything else on the page — running headers, page numbers, journal names, footnote fragments, and two-column layouts read straight across the columns into word salad. On a normal screen your eyes route around that clutter without noticing. In an RSVP stream there is no routing around anything: every extracted token arrives at the focus point, one at a time, with equal authority.

What the import pipeline actually does

Velo treats PDF import as a structuring problem and solves it on-device before a single word is streamed:

• Chapter detection — chapters, parts, and sections recovered from the document's structure, with a hierarchical table of contents and per-chapter progress.
• Running headers and page numbers stripped — the difference between reading a paper and reading the words "Journal of Cognitive Science" forty times.
• Two-column layouts resequenced — academic papers read in actual reading order.
• Footnotes and endnotes extracted and linked — matched to their reference points instead of erupting mid-sentence in the stream.
• OCR artifacts cleaned — and scanned documents go through on-device Vision OCR, with an optional server-side pass for genuinely difficult scans, metered in credits and invoked deliberately.

The result is that RSVP gets what it needs: clean, ordered prose. What you read is the author's words — nothing else.

Reading a PDF in the stream

Once imported, a PDF behaves like a real book. RSVP runs from 50 to 1,000 WPM with the optimal recognition point aligned to a fixed focus mark and timing that adapts to punctuation and word complexity. When you hit a dense passage — or a figure, a table, an equation, anything that's meaningless as a word stream — one tap drops you into Reader mode: full typeset pages, your position highlighted, eyes free to roam. Jump back and the stream resumes exactly where you left it. Scrub lets you drag backwards through the words you half-caught.

That mode-switching matters more for PDFs than for any other format, because technical and academic documents are exactly where a pure word stream breaks down. The stream carries you through the prose; the page view is there for everything the prose refers to. For the full picture of reading papers this way, see reading academic PDFs on an iPad.

The practical details

Import from Files, Safari, or Mail via the iOS Share Sheet. Imported PDFs live on-device and read fully offline. Both reading modes are free, along with import for up to 3 books, reading stats, and the intelligence dashboard; Pro removes the library cap and adds cloud sync across iPhone, iPad, and Mac.

If your PDF reading is mostly 40-page reports and papers rather than novels, this is the use case Velo's pipeline was built around — and the one where the difference between "streams the file" and "structures the file" is most visible.

← All guides