IBM to unveil speech recognition computer tech for Indian masses
Wednesday, 09 October 2002, 19:30 IST
Printer Print Email Email
NEW DELHI: You don't have to be a Gen-X tech wizard to imagine that computers of the future will listen to your spoken commands - that too in Indian languages. If the global tech giant IBM has its way, then one can soon open a Web site or spreadsheet page on the computer screen without tapping on the keyboard. All one has to do is speak out the command in Hindi or any Indian language. "We expect to launch the speech recognition tool in the Indian market in the next few months," said Manoj Kumar, director of IBM India Research Laboratory, the eighth research laboratory of IBM Research. Three of the seven IBM research laboratories are located in the U.S. and four are situated in Zurich, Beijing and Tokyo and Haifa, Israel. "Initially we are developing only Hindi-enabled speech recognition technology. Developing other Indian languages would be a lot more easier once we successfully launch the Hindi-enabled tool," Kumar told IANS. According to experts, speech recognition computer technology can be very useful in the Indian context as it provides an easy interface for interacting with computers for those unfamiliar with computers and English. Using such a convenient means of rendering information to or from the machine would mean that the end-user need not be computer literate and still can use the power of the IT industry. "Since there is no standard for keyboard input in Indian languages, using speech recognition for such languages would eliminate the need to know the keyboard mapping for the different fonts and related problems," said Kumar. He said that speech recognition technology could be a vital step in bringing the IT world to the vast non-English speaking Indian masses. "Currently we are working on increasing the accuracy of our Hindi speech recognition system by enhancing the acoustic and language models. The recognition system performs well over a trained context. "Our goal is to cover more Indian languages and then to build a multilingual speech recogniser for the Indian languages based on a multilingual phone set." With advances in speech-enabling technologies, proliferating electronic devices and use of the Internet, the market for speech and language technology is likely to grow from $1.4 billion in 2000 to nearly $3 billion by 2005, according to IBM. On the development aspect of Indian languages-enabled speech recognition tool, Kumar said: "Building a recognition system that would help a machine understand and transcribe the human speech has been a challenging field of research. "For any given language, the computer first needs to learn the sounds of spellings in various contexts from a huge corpus of speech and the corresponding transcribed text." Set up in April 1998, the IBM India Research Laboratory, located inside the sprawling campus of the Indian Institute of Technology here, is working on design and prototype of the e-governance solution, Middleware. Middleware will enable governments to deploy applications quickly at low cost, said Kumar, adding the solution can be customised to address the unique needs of individual state and local governments. "Using Middleware, government processes can be updated automatically without requiring substantial reprogramming effort and new policies can be specified in an instructive language by a non-IT person," he said, refusing to disclose the timeframe for the product's launch. Kumar said that the Indian research wing of IBM was also doing significant research and development activities in the emerging field of bio-informatics. "It is one of the biggest areas of our research activities."
Source: IANS
siliconindia