IBM Exploring Voice-Driven Web For the Developing World

Accessing the Web has become so easy for so many that it’s startling to learn that less than a quarter of the global population can connect to the Internet, according to Internet World Stats. Meanwhile, the penetration of mobile phones—especially in developing countries—is much larger.

Which is why IBM Research India in New Dehli has launched what it calls the Spoken Web, a network designed to use phones, not computers, to bring information on the Web to mainly under-served populations that cannot readily experience the Web’s benefits.

The Spoken Web is an exploratory project designed to bridge the digital divide. "A user can call up the system on a phone the way they would make any kind of call, interact with it for a few minutes, answer a few queries, supply some information that is asked by the system, and the result is what we call a Voice Site that is deployed against a specific phone number," says Arun Kumar, the IBM researcher who originated the project. "That phone number can be compared to a website’s URL and anyone calling that number can navigate the Voice Site just as they would a website—only by using voice instead of a visual interface."

While the Spoken Web relies on "programming by voice," Kumar says, its programming is not equivalent to professionals writing code line-by-line.

The difference "depends on where you draw the line between programming and configuring," says Robert Capra, a research scientist in the School of Information and Library Science at the University of North Carolina at Chapel Hill, and a specialist in speech and language processing. "My understanding is that they are working at a high level; they are using a phone and, with no other display other than voice input and output, they are allowing linking, providing conditional personalization, setting up calendar and appointment scheduling options, and are providing appropriate templates within the application."

"I have a lot of respect for what they are doing," adds Capra, "which is making it possible to support Internet-age business activities for a population that doesn’t have access to personal computers."

Spoken Web is still in the development stage and several pilot programs have proven its feasibility, says Kumar.

But there are still hurdles, especially in convincing a non-technical and, in some cases, illiterate population to overcome a bias towards technology.

"It takes a good amount of effort to get people to understand that talking to a phone for a few minutes is a very good alternative to walking a few miles or spending a day gathering information," says Kumar, contrasting the Spoken Web to an approach common in rural India.

In terms of speech recognition, India has 25 official languages, each with several dialects. "We have bypassed that problem as much as possible by not relying too heavily on speech recognition," Kumar adds. "We limit it so that it’s very controlled and is only used to help interact with the system."

He would not hazard a guess as to when IBM will make the system available for commercial applications.

Paul Hyman was editor-in-chief of several high-tech publications at CMP Media, including Electronic Buyers’ News.