Tachi

How to Speed Read a PDF or EPUB with an RSVP Reader

To speed read a PDF, import it into an RSVP reader, which extracts the text and shows it one word or short phrase at a time at a speed you choose. The PDF must contain real text: scanned, image-only files need OCR first. Start near 250 to 300 words per minute with one to three words per chunk and raise the speed only while you still follow the meaning. Layout, tables, footnotes and equations do not survive, so the method suits linear prose.

8 min read · Published

By Tachi, an independent project by Emin Berkay Dağlar. Editorial standards

To speed read a PDF, you need a reader that pulls the text out of the file and plays it back at a pace you set, one word or short phrase at a time. That method is called RSVP, and it works on any PDF that contains real text. The steps are short: check that the PDF has a text layer, import it, choose a speed and chunk size that suit the material, and stop when you stop understanding. The same steps apply to EPUB ebooks.

What you get is steady pacing and fewer distractions, which helps with long, linear prose. What you do not get is the page: tables, figures, footnotes and equations do not survive, and you give up the easy backward glance that normal reading allows.

Step 1: Check that your PDF contains real text

A PDF can hold text in two ways. A PDF exported from a word processor, a web page or a journal site stores actual characters, known as the text layer. A scanned PDF stores a photograph of each page. It looks the same to you, but to software it is a stack of images with no letters in it, so no text tool can read it.

A ten-second test tells you which kind you have:

  1. Open the PDF in any viewer and try to select a sentence with your cursor.
  2. If individual words highlight, there is a text layer. Copy a line and paste it somewhere to confirm that the characters come out correctly.
  3. If you can only draw a box over the page, or the search command finds nothing, the file is image-only.

An image-only PDF has to go through optical character recognition (OCR) first, which converts pictures of letters into text. Many PDF editors and scanning apps can do this and save a new, searchable copy. Check the result before a long read: OCR makes mistakes on poor scans, small print and unusual fonts.

Password-protected PDFs cannot be read by text tools either until you save an unprotected copy, and copy-protected (DRM) ebooks cannot be read by third-party tools at all.

Step 2: Import the PDF or EPUB into an RSVP reader

Tachi's reader is free, needs no account and runs in the browser. It accepts PDF, EPUB and plain-text files (TXT and Markdown), or text pasted from the clipboard.

  1. Open the readerand press the + button, then choose "Upload file". On a computer you can also drag the file onto the page. Once your library has at least one text, an "Import" button appears beside "Your texts" as a shortcut.
  2. Wait while the file is read. Every page is processed on your device, so a long book takes a moment.
  3. Check the title. It defaults to the file name without its extension; an EPUB uses the title stored inside the book. If you came through the + button you can edit it before saving.
  4. Open the text from your library to start reading.

During import the reader tidies the usual PDF damage:

  • Words hyphenated across a line break are joined back together.
  • Lines broken mid-sentence are merged into continuous text, while blank lines between paragraphs and pages are kept.
  • EPUB chapters are read in the book's own reading order.

If the import fails

Import error messages in Tachi and what to do about them
MessageWhat it meansWhat to do
Couldn't read this PDF. The file may be encrypted, image-only, or corrupted.The PDF would not open or has no text layer.Run OCR on a scan, or save an unprotected copy, and import that.
That file is empty.The file opened but contained no text.Check the file in another viewer; export it again if needed.
Couldn't read that file.The format is not supported, or the file is damaged.Convert it to PDF, EPUB or TXT, or paste the text instead.
This text is too large to save in this browser. Delete some texts and try again.The browser's storage for the site is full.Delete texts you have finished, or split the book and import one part at a time.

On the last point: only the extracted text is stored, so what counts is the amount of text, not the size of the file. A large PDF full of images may fit easily, while a very long book may not.

Step 3: Choose a mode, speed and chunk size for the material

The reader runs from 100 to 1000 words per minute in steps of 25, starting at 300, and shows 1 to 7 words at a time. For reference, adults read English non-fiction silently at about 238 words per minute on average (Brysbaert, 2019), so the default is already a brisk pace. Reasonable starting points:

Suggested reader settings by type of material
MaterialModeSpeed (WPM)Words per chunk
Light articles, newsletters, easy fictionFlash300–4001–2
Reports and general non-fictionFlash or Fade250–3252–3
Textbooks, papers, anything denseHighlight or Pacer200–2753–4
Text in a language you are learningHighlight100–2001–2

These are suggestions to adjust, not targets. Flash and Fade show one chunk at a time at a fixed point. Highlight and Pacer keep the whole passage on screen and move a marker through it, which preserves sentences and paragraphs and suits harder text. The differences are covered in chunking and on the How Tachi works page.

Then tune it with a simple loop:

  1. Read for two minutes at the starting speed.
  2. Pause and say, in one or two sentences, what you just read. If you can, raise the speed by 25 WPM.
  3. If you cannot, step back, drop 50 WPM and stay there. That is your working speed for this document today.

Leave "Pause on punctuation" switched on: it holds chunks that end in a comma a little longer and chunks that end a sentence longer still, which gives you time to close each thought. "Smart chunk" groups words by length so long words are not crowded. On a keyboard, Space pauses, the left and right arrows step one chunk back or forward, and the up and down arrows change the speed.

Step 4: Read in sessions and pick up where you stopped

Paced reading takes more sustained attention than ordinary reading, because the text does not wait for you. Sessions of 10 to 20 minutes with a short break between them are easier to keep up than an hour in one go.

Your position is saved as you read and when you close the reader. The text you were last reading appears under "Continue reading" with its progress, and opening any unfinished text resumes from the saved word. You can drag the progress bar to move to another place, or choose "Start over" in the options. Positions are stored in the browser, so they do not follow you to another device.

Does your PDF stay private?

In Tachi, yes. The file is opened and converted to text by code running in your browser; it is not uploaded. The resulting text, your settings and your reading positions are kept in that browser's local storage, and there is no account or server copy. Deleting a text, choosing "Reset all data" in Settings or clearing the site's data removes it. The details are in the privacy policy.

This is worth checking with any tool. Many online PDF converters and readers work by uploading your document to their servers, which matters if the file is a contract, a medical letter or unpublished work.

What RSVP cannot do with a PDF

  • Layout is lost. Tables arrive as a run of cell contents, and text in multiple columns or sidebars can come out in an odd order.
  • Figures and images are dropped. Only their captions, if they are text, come through.
  • Equations and code break. Notation that depends on position, such as fractions, superscripts and indentation, does not survive being flattened into a line of words.
  • Page furniture is included. Running headers, page numbers and footnotes are text on the page, so they turn up in the middle of the flow.
  • Looking back takes an action.In normal reading your eyes jump back when a sentence did not make sense, and experiments show this supports comprehension (Schotter, Tran & Rayner, 2014). With one chunk on screen you have to pause and step back instead. See regression in reading.

RSVP suits linear prose where following the thread is enough. For anything you must study, check or sign, read the original PDF at your normal pace.

Nor will the method multiply your speed for free. The research review by Rayner and colleagues (2016) concludes that eye movements take up only a small share of reading time, so removing them saves little, and that pushing well past your normal rate costs comprehension. The honest case for speed reading a PDF is focus, a steady pace through a backlog and modest gains, as set out in does speed reading work?

Other ways to get through a PDF faster

  • Skim it first. Read the abstract or introduction, the headings and the conclusion, then decide which sections deserve a full read. Of all speed-reading behaviours, selective reading has the best support; see skimming vs scanning.
  • Search it. If you have a specific question, the find command gets you to the answer faster than any reading method.
  • Read the important parts only. Copy the two or three sections that matter and paste them into the reader as a separate text.
  • Fix the display. A single-page view zoomed so the text column fills the window, with a comfortable line length, removes a good deal of friction from on-screen reading.
  • Prefer EPUB when you have the choice. EPUB stores text as text in a defined reading order, so it converts more cleanly than a PDF of the same book.

Frequently asked questions

Why can't an RSVP reader open my scanned PDF?

A scanned PDF is a set of page photographs with no characters stored in it, so there is nothing for a text tool to extract. Run optical character recognition on the file first, using a PDF editor or scanning app, and save the searchable copy. Check the recognised text for errors before starting a long read.

Is my PDF uploaded when I import it into Tachi?

No. Tachi opens the file and extracts the text inside your browser, and the file is never sent to a server. The extracted text is saved in your browser's local storage along with your reading position. There is no account, and clearing the site's data or using Reset all data in Settings removes everything.

What speed should I use to speed read a PDF?

Start between 250 and 300 words per minute, which is slightly above the average adult rate for non-fiction, and go up 25 at a time while you can still summarise what you just read. Drop lower for dense or technical material. A speed at which you lose the thread saves no time, because you will have to read the text again.

Can I speed read a textbook or a research paper this way?

Only the prose parts, and with care. Tables, figures, equations, footnotes and multi-column layouts do not convert well to a stream of words, and dense arguments often need re-reading. A better approach is to skim the document first, read the difficult sections normally, and use a paced reader for the long descriptive passages.

Can I import EPUB ebooks as well as PDFs?

Yes. Tachi reads EPUB files in the book's own chapter order and uses the title stored in the book. EPUB usually converts more cleanly than PDF because it has no fixed page layout. Books protected by DRM, such as most purchases from large ebook stores, cannot be read by third-party tools.

Sources