Advertisment

We need acoustic models in Indian languages 

author-image
DQW Bureau
New Update



Advertisment

The `Indo-European Conference on Multilingual Communication Technologies’
organized by Center for Development of Advanced Computing (C-DAC) at Pune saw
industry experts underscoring the need for creating acoustic models in Indian
languages for practical use.

Dr S Ramani, director, H P Labs, India who delivered the keynote address on
`Speech Interfaces in Indian languages’ was of the firm opinion that automatic
speech recognition and text to speech in Indian languages was yet to come of
age. “We can create applications using tools available in the market, but good
local language acoustics are necessary for practical use,” he explained.

According to Dr Ramani, English continues to remain the main barrier for
access for IT tools. Most IT users in India use English and what is available to
the others is not adequate, he said. Just about five percent of the total Indian
population reads English newspapers today. For the 1.8 million PCs sold every
year, only 40,000 copies of Indian language software are sold.

Advertisment

Barely one per cent use Indian language software, he pointed out.

The real problem, according to Ramani, is that most IT tools were originally
designed for sophisticated users. New simple designs are necessary for usage to
spread to billions of people and to overcome complexities, he said, underlining
the need the address the needs of the illiterate as well. How far can speech
interfaces go? Only a few special applications like finding train timings, or a
variety of important applications that every user can exploit. Simple mode of
speech communication is necessary, he said, adding that only speech interfaces
offer the hope of simplicity.

The problem with speech interfaces is that the technology is not robust
enough for widespread use. Moreover, special fonts have to be made by interface
design to cope with the ambiguity in speech, he said, adding that there are deep
problems in correct recognition of speech.

Advertisment

Rajeeva Ratna Shah, secretary, DIT, was of the opinion that India’s major
problem was the growing economic as well as digital divide. He called for
applications that were low cost, voice activated, user- friendly and cheap as
well.

He also called for well-directed and focused efforts to do language
programming. If we can resolve the language issue, we can become major language
software providers to the world, he said. In this context, he called upon C-DAC
to transcend beyond fonts and keyboards and move onto areas such as word
processing, office software and text to speech software.

With more than 40 odd institutions working in language software, the
government has begun a special initiative in association with C-DAC, NCST, TDIL
and Vishwa Bharati to ensure that these work together in a connected manner and
there is no duplication in their efforts. RK Arora, ED, C-DAC, welcomed the
dignitaries.

Advertisment

Nanda Kasabe

CNS

Advertisment