Can't Read? Talking Phones Will Help

[Chirag 5]

Can't read? Talking phones will help

Hemangi Balse in Mumbai | April 24, 2004 09:21 IST

Rediff.com

Illiterate but want to use a mobile phone? Not to worry -- help is at hand, courtesy Reliance Infocomm. The Ambani company is launching speech-based applications. Quite simply, that means that an illiterate villager or urban dweller has merely to mention the name of the person he or she wants to talk to and, presto, the mobile phone will do the rest of the job -- by scanning a Reliance phone directory.

More to the point, an illiterate person can speak in any of several Indian languages and dialects. So a villager in the western ghats of Maharashtra (who speaks a different Marathi dialect) can speak in his dialect into the phone, without pressing too many buttons.

Nor does he have to buy a new mobile phone. Says Mahesh Prasad, president, applications and solutions group, Reliance Infocomm: "We are working on a network-based technology that is independent of handsets. This will mean that the current handsets can be used to recognise and transmit voice signals."

Prasad declines to say when the service will be launched, but others expect it to be launched in a few months.

How do these voice-based commands work? The handset will recognise the command, translate it into text, retrieve the phone number of the person being telephoned from the data base, translate the data into voice and transmit it to the user's handset.

Globally, much work has been done to recognising speech in English (in various dialects). However, the biggest challenge is to offer this in various Indian languages with several dialects, besides commercialising these services, says Prasad.

Reliance Infocomm has already set up a dedicated team to look at the permutations and combinations of speech-based applications and products for customers.

According to the 2001 census, 34.62 per cent of India's population is illiterate. Reliance's current effort is clearly aimed at expanding the mobile services market.

Reliance is working on several "futuristic applications" for its R-World. It will soon be launching an application to make a mobile phone more accessible to the blind. This application, for which Tandem Infotech received a Dhirubhai Ambani Developers Programme award, enables a visually impaired person to "hear" missed calls or an SMS (which is automatically read out to him) and dial back the number.

Though the application is meant to directly benefit the blind, it is also expected to be a high utility tool for anyone using a mobile phone.

[Greens 21]

Funny ppl, why does an illiterate who cant even recogonise numbers want a mobile phone what difference is it going to make?????

[Ashokjp 15]

For some people it does make a difference..

Maybe trend or even convinience...

Some ppl may have enough money but maynot be literate

Cells are coming to be a need in everyones life..

So it helps them a lot when they buy cell..

Not only this, there are many other instances too.

[city02 63]

Language is perhaps the most important single characteristic that distinguishes human beings from other animal species. The vocal apparatus of humans

has the ability to produce a wide variety of sounds, and our ability to handle grammar and linguistic complexity is also phenomenal. In this sense, spoken language can be viewed as a species-specific characteristic of humans.

There are presently over 6000 tongues spoken around the world, with distinctions ranging from regional dialects of Hindi, to great language families like the

Indo-European and Sino-Tibetan.

Not only do languages have their own set of words, they often have unique phonemes and sentence constructions. For example, Indian languages have

retroflex consonants (sounds produced by curling the tip of the tongue under the palate). Indian sentences have a subject-object-verb (SOV) word order, as opposed to the subject-verb-object (SVO) word order used in English. So

'cats drink milk' would be ordered as 'cats milk drink' in Indian languages (e.g., billi doodh peeti hai).

Furthermore, Dravidian languages of southern India and Indo-European languages of northern India have numerous borrowed words and phonemes. For example,

Tamil has many loan words from Sanskrit, such as padam or foot. More recently, the advent of English in India has given rise to new language mixtures such as

'Hinglish', which is a cocktail of Hindi and English.

Strictly speaking, the English as spoken by Indians sounds different from that spoken in the UK or the US.

Phoneticians and linguists therefore qualify the English spoken in India as Indian English.

Due to recent advancements in speech technologies, machines can now recognize spoken words and sentences (speech recognition); synthesize words and sentences from text (speech synthesis); and even identify speakers (speaker recognition), with high fidelity. Machines that can interact with people using

these speech technologies are now being put to use for all sorts of services. These include, voice portals, voice dialing, automated triaging of calls at call centers,

automated help desks, dictation, and more recently control of automotive functions such as AM/FM radio and car windows, using speech.

These commercial services are now supported in many European languages. However, there is precious little that has been accomplished in building such systems for Indian languages (including Indian English). The diversity of Indian speech at the acoustic, phonetic and linguistic level; and the various dialects that exist in India make this a challenging task. There are several fundamental steps that need to be taken before speech systems and services can be built for Indian languages.

The first is to build a corpus of speech in Indian languages that is balanced from a phonetic, linguistic, and acoustic perspective.