MathSpeak, MathML Cloud and Accessible Math

Looking back

The modern conventions for reading mathematical notation out loud have evolved over time, motivated by the desire to improve the quality of mathematical instructional materials. Abraham Nemeth publicly shared his MathSpeak guidelines (1995), which have since been further expanded and refined through the MathSpeak Initiative (2004).

Modern text-to-speech (TTS) systems use various rules to determine which speech sounds are generated for a string of electronic text. Sophisticated TTS applications may offer more than one speech rule for the same string of text.

When using MathSpeak, the speech being generated closely matches the Nemeth braille code, leaving it to the reader (player) to interpret the semantics. However, people with other types of print disabilities (e.g. dyslexia) may find the MathSpeak speech style inaccessible.

The same Mathematical Markup Language (MathML) content can be used to generate a variety of speech styles without applying anything extra to the content. Steve Noble created examples with audio clips to demonstrate this for the University of Oregon's MeTRC website. They can be found at:

Additional information is provided on the University of Washington website.

MathML Cloud

Benetech built MathML Cloud to address the challenge of making digitized math accessible across the widest range of learners. The challenge stems from the fact that most standard reading technologies lack native support for MathML, an open standard for encoding mathematical expressions in web pages and digital textbooks. That is why content developers often default to rendering digitized math as an image. Sometimes they do so without retaining the detailed math markup for reference. In the absence of this information, however, when a screen reader user comes across the image, s/he hears the word “image” spoken, but has no means to perceive the math content behind it. More information is available on the Bookshare blog.

See also: