Speech recognition software enables one to not only create documents by dictating to the computer, but also run programmes and even make web pages through voice commands.

"The keyboard is one of the biggest inhibitors to the use of computers. Speech recognition will make computers not only easier to use but also accessible to more people," says Amar Babu of HCL Infosystems, which is marketing speech recognition products from an American major in the country.

Research is going a step further to make using a computer no different from speaking to a friend, says JR Isaac, an advisor with NIIT working on developing special software for the handicapped.

"Now we are trying to build cognition into the computer to make it more human and able to recognise the user," says Isaac.

"While today one has to loginto a computer and type one's password, we are working on systems with which you just say `hello' and the computer will recognise you by your voice and become active."

Speech recognition is an advancement on the voice recognition technology used in high-security systems in the 1970s.

While voice recognition distinguished between the voices of different people, speech recognition is an `intelligent' software which can identify spoken words and can be trained to follow different accents, says Sudip Dasbiswas of HCL.

The software, which comes with an in-built vocabulary of up to 200,000 words, works by matching waveforms of the user's speech with his speech file, created during an initial `training' in which the user reads out a pre-determined text to the computer.

The speech file records the accent, loudness, pitch, pause between words and other aspects of the user's speech, and the software matches the user's speech with this file to identify the words spoken.

The software keeps learning with eachsubsequent use, adjusting itself to the user's speech to reach accuracy levels of up to 98 per cent in recognising the words spoken, says Dasbiswas.

"The software attunes itself to the user's speech to such an extent that if the user is a lawyer who frequently uses a term like adjourned, it will type it on the screen even before he has pronounced the whole word.

The user can also edit and format a document through voice by `telling' the cursor to go to menu items, click on them, select words, lines or paragraphs, and so on.

And if the user wants to go over a document without having to read it, a read option enables him to have it read out by the computer in a synthesised voice or the user's voice if he so desires.

And the technology enables one to do more than just create documents. High-end versions of speech recognition software allow users to run any windows-based application by voice, says Babu.

For accuracy, speech recognition software requires a special microphone which filters out allsurrounding noise so that the speaker's voice is registered clearly on the computer.

But that may not be required in the future with the microphones being built into the computers, says Babu.

Recognising the potential of speech recognition technology, software firms are working to make it more user friendly.

"Intel and Microsoft have identified speech recognition as the future. In fact Intel's Pentium III processor is geared for such a technology," says Babu.

One limitation with speech recognition software in the market is that it cannot be used on local area networks. The way the software is built, it allows only one user to use it at a time.

But this too will soon be overcome, says J Risaac of NIIT, who is developing special software for the handicapped.

And speech recognition will soon expand to electronic gadgets, says Babu. "We will be able to switch things on and off, switch channels on television and do with voice commands everything that we do currently do through knobs, switches or remotecontrols."

With processing power increasing exponentially year by year, PCs with such interactive capability will be in the market in two to three years, according to Isaac.

But speech recognition is just the beginning of interactivity; researchers are already working on operating computers by thought.

The technology, still at the laboratory stage, enables a computer to execute a thought command from a user by analysing the brain wave patterns of the thought which it picks up through electrodes attached to the user's body.

"The challenge," says Isaac, "is to build into a computer all the five senses which humans possess."