InftyReader

*InftyReader AI Version (Online)

We have launched the online version of InftyReader AI : https://inftyreader.online.

On this site, you can upload a PDF and download the recognition results produced by a Large Language Model (LLM)–based generative AI in the following data formats:

• LaTeX (.tex)
• MS Word document (.docx)
• Text (.txt) with human‑readable TeX notation for math
• Markdown (.md) with LaTeX math notation
• HTML (.html) with MathML
• EPUB3 (.epub) with MathML

A free trial site is available, so please feel free to try it: https://inftyreader.online/trial.

The trial site can be used for up to 300 pages.

If you wish to continue using the service beyond that, please click the Purchase button below and buy a license code.

* Purchase with Stripe (InftyReader Online License code)

When you make a purchase, an 8‑character alphanumeric license code will be downloaded. Please note that the license code is case‑sensitive.

How to use it:

Visit the InftyReader Online website at https://inftyreader.online, enter your purchased license code in the License code field, and click the “Go” button.

Then, your available page count will be displayed in the menu bar.

Each time you run a recognition process, the available page count will decrease.

When the count reaches zero, please purchase a new license code.

---

The traditional Windows application version of InftyReader based on OCR and PDF parser technology can be downloaded from below:

---

InftyReader (Windows application)

InftyReader is OCR software to recognize scientific documents including mathematical formulae (STEM documents).

• "InftyReader" converts PDF and scanned images to various types of accessible documents: LaTeX, XHTML(MathML), HRTeX, IML, Microsoft Word document, EPUB3 and PDF with TeX.
• For the scanned image files or Image PDF produced from scanned images, InftyReader uses OCR specially trained for STEM documents recognizing special math symbols and analyzing math structures.
• For e-born PDF(*), InftyReader uses a PDF parser rather than OCR, so the character recognition results are very accurate, not only for ordinary texts but also for math symbols.

(*) e-born PDF is the PDF produced by authoring tools such as LaTeX system, MS Word, Adobe InDesign, etc. The PDF produced from image files are called Image PDF.

*InftyReader Ver.3.3.4.0 (June 1st, 2026)

Personal Use License package:
InftyReader3340.zip (English Edition, about 206MB) --- June 1st, 2026

Enterprise License package:
InftyReader3340_Enterprise.zip(English Edition, about 206MB) --- June 1st, 2026

What's new (Ver.3.3.4.0):

• Fixed the issue where the baseline of math formulas and text lines was misaligned.
• Improved compatibility with PDFs containing fonts that assign non-standard codes to math symbols.

For general information about InftyReader, please read "AboutInftyReaderE.txt" here.

What's the difference between the Enterprise edition and the personal use edition? Please read: "About InftyReader Enterprise".

License Update. The serial numbers of InftyReader ver.3.1 and 3.2 are valid for Ver.3.3, so all the users of InftyReader ver.3.1 and 3.2 series can use the version 3.3 series without any additional cost.

Trial Use. To use InftyReader in the Trial Mode, please see AboutTrialUse.txt.

Remark. Please avoid using the file names and the path names including NON-ASCII characters as the input file for InftyReader.

* Purchase with Stripe

Notice of Temporary Suspension of Software Sales

As of October 22, we have temporarily suspended the sale of new software licenses below as a precautionary measure to ensure the security of our customers' information and services while emergency maintenance is being carried out.

This action has been taken in response to confirmed reports of a critical vulnerability affecting software used on our servers. At this time, we have found no evidence of unauthorized access, data breaches, or any compromise of customer information.

We sincerely apologize for the inconvenience this may cause and appreciate your understanding and patience while we work to address this issue.

We will announce the resumption of software sales as soon as a clear timeline has been established.

Customers who have already purchased software from this website and have successfully completed license activation are not expected to be affected at this time. However, customers who purchased a one-year license may be required to install a software update in the future. Any updated version of the software will be made available on this website. Once released, we kindly ask affected customers to download and install the update at their earliest convenience.

Thank you for your continued support and understanding.

You can get a serial number immediately after the payment on the purchase site above, and you can use it to activate InftyReader on your PC to start using InftyReader.

N.B. Please note that InftyReader DOES NOT accept LOW RESOLUTION image files. Please read the specifications written in AboutInftyReader. We recommend you to test InftyReader in trial mode before purchasing.

* Purchase with Paypal

In case you wish to purchase with PayPal, please visit here.

Please note that you will receive the serial number within 2 business days after the payment with PayPal.

N.B. Please note that InftyReader DOES NOT accept LOW RESOLUTION image files. Please read the specifications written in AboutInftyReader. We recommend you to test InftyReader in trial mode before purchasing.

Document for blind users.

Below is the Introduction to InftyReader for blind users given by Prof. John Gardner (Oregon State University & ViewPlus Technology) at the ICCHP Summer University 2011.

Introduction to InftyReader by Prof. John Gardner.

* Comments about output formats

1. IML is the default XML file format of the editor "InftyEditor", an authoring tool of math documents developed by InftyProject. InftyEditor provides a very easy user interface to input and edit math expressions together with ordinary texts.
   The English edition of InftyEditor is free software. Please see the sites of InftyEditor.
2. LaTeX is a widely used common markup language for writing mathematical documents among specialists in science.
   
3. In XHTML format, mathematical expressions are output using MathML notation.
   
4. HR-TeX is a simplified LaTeX-like notation easier "to read" specially designed for the blind.
5. Word XML output from InftyReader can be directly imported into Microsoft Word.
6. In EPUB3 format, mathematical expressions are output using MathML notation.
7. In PDF with TeX is a newly proposed Accessible PDF. Its front image is the same as the original PDF, and the text and math information is embedded behind the image rearranged in the usual reading order. Math expressions are embedded using HR-TeX (Human Readable TeX) notation. To get the "PDF with TeX" output, users are recommended to install Ghostscript. Its download site is below:
   https://www.ghostscript.com/download/gsdnld.html
   If you have Ghostscript installed on your PC, the front image of the output PDF will be the high quality vector image generated by Ghostscript.

* Caution ---- Important!

To recognize image files:

1. Source documents have to be clearly printed.
2. It should be scanned in 600dpi (or 400dpi) with no distortion. Usually, binary images are better for recognition than color images.
3. InftyReader erases small noises, segments page images into picture areas, table areas, and text areas automatically, and then recognizes text/table areas including mathematical expressions.
   However, to get better recognition results, users are <<recommended>> to erase noises and pictures before the recognition.
4. In scanning, it is important to adjust the binarization threshold of the scanner so that the number of touched or broken characters is less than 1% of the total number of characters in each scanned page image.
   

* Operating Environment

InftyReader runs on Windows 10 and 11, on a PC equipped with at least 2GB free memory available for the application.

* How to use InftyReader?

1. Select file(s) or folder.
2. Input/select output document name
3. Press the "Start" button.

Then, the recognition results of the selected image files are saved into the file you specified by the "output document name". When you select a folder instead of files, all the image files in the folder of the specified file type (TIFF/GIF/PNG/BMP/PDF) are recognized and the results are output into the files having the name(s) of the folders.

If you set to check to the "Search Sub Folders" item under the "Option" menu, InftyReader recognizes all the image files in the subfolders of the selected folder. For example, if you select the folder "foldertop" having the subfolder structure below,

1. foldertop
   |-- subfolder1
   |        |-- a.tif
   |        |-- b.tif
   |
   |-- subfolder2
            |-- c.tif
            |-- d.tif

and if you select the file type "IML" for the output file type, then, you will get the files "subfolder1.iml", "subfolder2.iml" in the folder "foldertop". The recognition results of a.tif and b.tif (resp. c.tif and d.tif) are saved in the file subfolder1.iml (resp. subfolder2.iml, respectively).

If you select LaTeX as output file type, you will get "subfolder1.tex", "subfolder2.tex", and it is similar for other file types HR-TeX and XHTML.

* License

InftyReader is usable under the following license agreement.

(1) You may not modify the software in any manner. You may not reverse engineer, decompile or disassemble the software.
(2) You may not sell the software without making a formal agreement with Science Accessibility Net.
You may distribute the software only free of charge, without modifying the zip package of the software.
(3) The author shall have no obligation to correct errors and inconveniences of the software.
(4) The author shall not be responsible for any loss and damage caused by the use of the software.
(5) The license is limited to personal use, including the case purchased by an institution for the specified user. Shared use by a small group member is also allowed. In the default setting, the number of pages recognizable by this license is limited to 10000 pages per month. In case an institution uses the software to service several clients or to digitize huge numbers of volumes, please use the enterprise version, reading the page here: About Enterprise License. For more details, please contact us.

* Report

Any report about the software will be welcome.

--------------------------------------
Non-Profit Organization
Science Accessibility Net (sAccessNet)
e-mail: support"at"sciaccess.net (Please replace "at" by @.)
URL: http://www.sciaccess.net/
--------------------------------------