Home / PDF speed reader
A PDF speed reader for the stack you have to get through.
Papers, textbook chapters, reports, standards documents — the reading nobody chooses. Import the PDF and read it one word at a time, with the running headers and page numbers taken out, entirely on your own device.
Open ReadKinetic and drop a .pdf onto it. No install, no account, and no upload — which is the part that matters when the file is an unpublished manuscript or a document your employer would rather not see on someone else's server.
PDFs are harder than they look
A PDF does not contain paragraphs. It contains instructions for placing glyphs at coordinates on a page. Everything you think of as structure — sentences, columns, where the footnote ends and the body resumes — has to be reconstructed. That reconstruction is most of the work, and it is where most tools hand you a mess.
Running headers and footers are removed
The journal name, the chapter title, the author's surname, the page number — all of them repeat on every page, and all of them would otherwise arrive mid-sentence every few hundred words. Repetition is what gives them away, so they are found and stripped.
Hyphenated words are rejoined
A word split across a line break is put back together, while genuine compounds like "well-being" are left as they are. Getting this wrong in either direction is obvious the moment you read a page of it.
Scanned files are called out, not silently imported
If a PDF turns out to be page images with no text layer, ReadKinetic says so and points you at OCR — instead of importing an empty book and leaving you to work out why. If only part of it is scanned, which happens constantly with older papers, it tells you how many pages are affected.
The pace eases off where the text gets heavy
The reader slows slightly before words that are rare or long relative to the rest of this document, and carries a slower pace through passages that are denser than the document's own average — then lifts again. It is computed from the text itself, so a pharmacology chapter and a history chapter are treated differently with nothing to configure.
Your place is kept in long documents
Close a 900-page textbook mid-chapter and it reopens exactly there, weeks later, with no sign-in. Each document keeps its own reading speed, because the pace you read a review article at is not the pace you read a methods section at.
Nothing above requires a network. The PDF is parsed in your browser and the text is stored on your device, so it works on a plane, in a basement library, and in any situation where uploading the file would be a problem.
Read the paper in the right order first
The largest saving on academic reading is not speed, it is sequence. A paper is a structured document, not a story: read the abstract properly, then jump to the discussion — which tells you what the authors think it means — then look at the figures, and only then read the introduction and results quickly. Most papers can be dismissed in ninety seconds that way, and being done in ninety seconds is a win.
The full version of that, including which sections punish you for rushing them, is in How to read research papers and textbooks faster.
Formats it opens
If your reading is mostly books rather than papers, the EPUB page covers what changes — chapters, embedded fonts, and starting at chapter one.
What it will not do
It cannot read a scan. No text layer, no text. OCR first.
It cannot read equations, tables or figures. Those are the parts of a paper that carry the most information per square inch, and they need your eyes on the page, not a word at a time.
It will not fix a badly built PDF. Some files store their text in an order that has no relationship to reading order. Where that happens it will show, immediately, in the first paragraph.
Common questions
Can I see the actual pages, not just the text?
Yes. Every PDF can be read either as reflowed words or as the pages it was printed as, drawn from your own file, and you switch between them with one control. The page view is a real reader — figures, tables, columns and equations in place, two pages at a time on a wide screen, pinch or scroll to zoom into a diagram — and it is still part of the app: tapping a line starts reading from there, find searches the whole book and takes you to the page, and your marks and place work exactly as they do everywhere else.
Does it work on a 1000-page textbook?
Yes. Pages are drawn as they come near and released as they go, so a long book costs no more to scroll than a short one, and reopening puts you back on your page rather than travelling there. If the PDF carries its own table of contents it is used, so a large textbook opens with its real chapters rather than a list of page numbers.
Is my PDF uploaded to a server?
No. The file is parsed and rendered by JavaScript running in your own browser, and both the text and the file itself stay on your device. Nothing is sent anywhere at any point, including when the pages are drawn. This matters more for PDFs than for anything else, because PDFs are where people keep unpublished papers, work documents and course material.
Why does my scanned PDF come out empty?
Because it has no text in it. A scanned document is a stack of page images, and there is nothing for any reader to extract. ReadKinetic detects this and tells you outright rather than importing a blank book — run the file through OCR first (Acrobat, macOS Preview, or any OCR tool) and import the result.
What happens to page numbers and running headers?
They are stripped. A running header repeats on every page, so it is detectable by repetition and removed before the text reaches you — otherwise the journal name would land in the middle of a sentence roughly every 400 words.
Does it handle two-column academic papers?
Yes, in two different ways. Read as words, it follows the order the PDF itself declares, which is usually correct — though some papers store their columns in an order that does not match how you would read them, and then the text interleaves. Read as pages, the question does not arise: you are looking at the actual typeset page, columns where the typesetter put them, and tapping any line starts reading from there. On a wide screen you get two pages side by side.
Can it read the equations?
It cannot say them one word at a time, and neither can any word-at-a-time reader — mathematical notation is not made of words. But it can show them. Switch to the original pages and the equations, diagrams and tables are exactly as printed, so the usual pattern works: take the prose at speed, stop at the mathematics, and read that at the speed of the reasoning. That is a real reader and a speed reader in one place, which is the point.
What speed should I use for a research paper?
Not one speed. Introductions and discussion prose take 400–500 words per minute comfortably; a first pass through an unfamiliar argument wants 250–350; methods, definitions and anything you will be examined on need the slow setting or a different approach entirely. Switching between them is the actual skill.
Is it free?
Yes. No account, no trial, no page limit, no paid tier. Optional donations and a small ad on the blog are the whole business model.
What other formats can it read?
EPUB, Kindle books (MOBI, AZW3, PRC), FictionBook (FB2), Word (DOC, DOCX, DOTX), OpenDocument (ODT), PowerPoint (PPTX), AbiWord, RTF, HTML, Markdown, LaTeX, reStructuredText, Org-mode and plain text — plus pasted text. EPUB has its own page.
Open a file and see.
Free, no account, nothing uploaded. Works offline once loaded.
Open ReadKinetic →Related: Is my PDF scanned or readable? · EPUB speed reader · How to read research papers faster · Speed reading for students · How to read your own books faster · Arabic and Hebrew speed reader · Anticipatory pacing