T.V. Raman’s Audio Deja Vu: From Google, a Math-Reading System for the Web
Almost exactly 20 years ago, T.V. Raman built a system for reading math out loud as his PhD project at Cornell.
After being raised in Pune, India and becoming blind at age 14, Raman had learned to program with someone standing behind him to read the display. He wanted to help other people gain access to the world of math by giving them an audio interface that could render complex technical documents that they could read for themselves.
So Raman built it himself — and named the math-reading project Aster, after his first seeing-eye dog.
But then Raman put Aster (the project, not the dog) aside, given that there was little electronic technical content available. “The software existed, but the content was lacking,” Raman recalled.
Today, Raman is a Google research scientist on the company’s accessibility team (though he prefers the broader term “usability”). And one of his latest projects is to bring math to the Web by building it into the screen-reading browser extension ChromeVox, which is also used in Android.
“Fast-forward to today: Now that we have the content and the distribution channel, math is finally becoming real on the Web,” Raman said.
It seems almost impossible that someone could grasp a complex equation just by hearing it read out loud, but the key aspect of Raman’s interface is that a user can traverse a technical equation back and forth and top to bottom within the logic structure, making sense of the broader picture as well as the small details.
As Raman described it in his thesis paper from 1994, “The passive nature of listening prohibits multiple views; it is impossible to first obtain a high-level view and then ‘look’ at portions of the information in detail.” But his implementation of math reading, past and present, changes that by allowing users to “browse” an equation.
Google’s new math-reading tool includes pitch changes and speech rate variations in order to designate things like superscripts and subscripts, just as Aster did.
Raman’s bit of audio deja vu isn’t quite ready yet; math reading is included in Chrome version 28, which Google only just promoted from the “dev channel” to the “beta channel.” That means it’s hard for the public to try this unless they want to sign up for a browser that’s not ready to be released. But Raman said he hopes to be “rock solid” by August or September.
Making math more accessible could open the door to studying the sciences for many blind students, who might never get a chance to even try math in their elementary school years, Raman said. He also has a suspicion that it could be very helpful for other kids who are non-visual learners.
“Math is intimidating because the notation is complex and cryptic — but what if someone is reading it to you?” Raman said.