Business

Speaking to computers

By A. A. Harichandan

An employee of Toshiba displays a prototype of the Bluetooth wireless communication headset, which enables a user to control computers by speech, at the company's laboratory in Tokyo. — AFP

BANGALORE APRIL 20. The Japanese electronics major, Toshiba, recently released its prototype of a light-weight wireless headset that could be used to `speak' to computers.

Time may come when Indian software companies, may be the ones building applications, to capture the services segment of the speech market.

The applications are many and the game has just begun. From simple queries like "when is the next train to Mumbai" to "how many disabled people live in Bangalore today who were immigrants in 1974?," when the speech market comes of age, all you have to do is, talk to the computer, television and phone, though they may all be the same device.

What is more exciting is typified by an experimental stage work in Bell Labs, U.S., a voice driven web browser. Right now, the software needs a powerful processor to work on, but factor in Moore's Law and any other relevant law, and some day we may really have it running on a cell phone. Indian companies are not looking at developing such products. They are, however, looking to cash in on the market one level down the ladder, the services market; here, they use the engines by built by others and build applications for clients like banks, hotels and so on.

The software and hardware that can recognise human voice or make sounds like us are produced by big players like IBM, Motorola, AT&T, Lucent, and specialists like Nuance Communications.

"Indian companies are not interested in spending millions of dollars on such projects, says N. Krishnamurthy, who runs VSM Software, a small speech applications software company here. Besides, once they make a synthesiser, they have to market it; and to quote Vivek Kulkarni, Karnataka's IT and BT Secretary, "We are weak in the product area, we prefer to go in for services". Mr. Krishnamurthy agrees: "It makes sense for us to use `speech engines' from IBM or Nuance and build applications for which we have a market; we don't have the resources to invest in making speech synthesisers ourselves".

In the long run, speaking computers represent cost savings for companies, say industry watchers. For a company hiring the services of a call centre, every call from a customer may cost between $1 and $10, depending on the duration as well as the skill of the agent. Analysts say, as many as 40 per cent of all calls are simple queries that computers can be programmed to handle. These calls could cost them as low as 20 cents.

There is potential for use in companies with large sales, distribution, service or customer networks, the railways, airlines and hotels, telecom service and Internet service providers, and the financial services companies. "We already have a project from a large American company," Mr. Krishnamurthy said. His company has also developed some pilot financial and directory services applications. They are talking to major telecom companies for collaborating in this area. Its not all foreign.

Not all work on voice is being done in the U.S. or Europe. Scientists here are quietly doing interesting work that has applications in hands-free dialing, a feature that many cell phones offer, to very specific intelligence coding for use by the defence services. For example, quality of voice, even when there is a lot of noise around is important for the hands-free feature. This has a market in many European countries where, using cell phones while driving is prohibited. Significant work in this area was done by Indian scientists, for Ericsson for example. For defence use, "Certain channels are available / used for short duration; we need to optimise, packing key words containing information into that time," said one scientist. "The words will make sense only to those who know the code. What we try is to improve the efficiency of the software that packs key words into the available time. The more we can work-in, in a limited number of bits, without compromising on intelligibility, the better.

Send this article to Friends by E-Mail

Business

News: Front Page | National | Southern States | Other States | International | Opinion | Business | Sport | Miscellaneous |
Advts: Classifieds | Employment | Obituary |

Stories in this Section

• Stability need of the hour
• Speaking to computers
• Forex reserves approaching $55b
• Building an environment for business growth

Copyright © 2002, The Hindu. Republication or redissemination of the contents of this screen are expressly prohibited without the written consent of The Hindu