Skip Ribbon Commands
Skip to main content

Interview with Prof. A. G. Ramakrishnan


Published on March 22, 2010


 

 

 Mr. A. G. Ramakrishnan received his Ph. D. from IIT, Madras in Biomedical   Engineering. After working with BPL India and Fetzer Institute, Kalamazoo,  USA, he joined Indian Institute of Science, where currently, he is a Professor  heading the Medical Intelligence and Language Engineering (MILE) Laboratory.  He received Thangam Vasudevan Award (Indian Association of  Biomedical Scientists), Sir Andrew Watt Kay Young Researcher's Award (Royal  College of Physicians and Surgeons, Glasgow) and Young Investigator  Award (DST, Govt of India). His research interests include Indian language  Technology Solutions and Image Processing in Medicine. He is a Senior Member of IEEE, Engineering in Medicine and Biology and Signal Processing societies. He was President of the Biomedical Engineering Society of India till Dec. 2008 He is a Fellow of the IETE and IE (India). He has guided 6 Ph.Ds, 12 Masters by research and 48 Masters in Engineering students. He is a Member of International Forum for Information Technology in Tamil, Project Advisory Committee, Linguistic Data Consortium for Indian Languages, and Project Review Committee for Biomedical Engg for Indian Council for Medical Research. Currently, he is directing a National level research project on “Recognition of online handwriting in Indian Languages.” His major commitment is to develop an “Automated Reader of Indian language printed books for the Visually Challenged.” MILE has already tied up with Bookshare.org, USA to make available digital books online in Tamil and Kannada.

How did Indian Institute of Science take interest in Indic Computing activities? When did it start?
About 2000. I saw that the token number announcement system in banks had very bad pronunciation, even though it handled only just 10 digits. I also realized that all around the world, people are working on languge technology for English only, both because resources are readily available and it is fashionable. Japanese, Chinese and even Indians were working only on speech recognition in English only. So, I decided to work on Indian languages only and not English. I also realized that Indian languages are having a slow death, and it is mainly the entertainment and the media, which is keeping our languages alive in the major cities of India. In the 63 years after independence, we as a nation have not done anything systematic to develop our languages in tune with the development of science and technology in the world. Thus, even if somebody wants to study in Indian languages, there are no books; there is no standardised vocabulary for technical words in any Indian language, including the national language Hindi. This is alarming, because our entire culture and literature will become unavailable for the future generations. In Japan, Germany and France, people can even write their Ph D thesis in their own languages. I wanted to really do something for this.
What are the challenges in Indic computing that are being addressed by Indian Institute of Science ? What are the achievements?
We, at MILE Lab, have good wroking prototypes in Tamil and Kannada OCRs as well as text to speech systems. Our text to speech system is available as a web demo at http://mile.ee.iisc.ernet.in/tts. A Tamil teacher from Singapore told me that she is using our TTS to teach pronunciation of Tamil words to their students and for their projects ! We have got feedback from over 100 people, and over 1000 people have tested the web demo around the globe. I am also working on online handwriting recognition in Tamil and Kannada. In collaboration with two private industries in Chennai and CDAC, Pune we have developed three form filling applications, where our handwriting recognition engines can be used to recognize the handwritten entries made by people. This being a new area of research in India, and also a tough one, currently these applications give only about 80% word level accuracy. Prof. N Balakrishnan, the Associate Director of Indian Institue of Science is working on machine translation involving Hindi and other Indian languages.
How did you get interested in recognition of online handwriting in Indian languages?
Working on printed OCR the interest naturally moved on to handwriting recognition, since it will also be a good natural interface for Indian languages to augment keyboard based inputting to the computers. People, who do not know English and are not used to the keyboard, might find handwriting input more comfortable and less daunting
What is the current status of handwriting recognition in Indian languages ?
I am leading a National level consortium research project funded by TDIL, MCIT, where partners from IIT Madras, IIIT Hyderabad, CDAC Pune and ISI Kolkata are working on Telugu, Malayalam, Hindi and Bangla, respectively. IISc is working on Tamil and Kannada. Learnfun Systems, Chennai, C K Technologies, Chennai and CDAC Pune have also developed Accident reporting form, encumbrance certificate application form and census data collection form, using the recognition engines developed by our scholars. Serious research started only 3 years ago, on consortium mode funding by TDIL, Govt. of Indiia. We have achieved about 85% accuracy on good handwriting, at the character level. Research in any language technology gets accelerated if we have good annotated databases. The current slow process is due to the lack of such databases. Hence, we have spent quite an amount of our energy in collecting good handwriting databases in all the above languages. It will easily take another 3 years to see the first application being used by people. Very soon, we are adding some Institutions as partners, who will be working on Assamese and Punjabi (Gurmukhi). I am also planning to conduct a National Level Technical Competition for college students to create novel, engaging applications and maybe, interesting games involving handwriting recognition. The vision is that the use of Indian languages on various computing devices, including handheld devices will increase significantly, with the right kind of application.
What are your other research projects with respect to Indian language computing?
Bilingual (Indian language and English) Text to speech synthesis and Printed text recognition (OCR) in Tamil and Kannada are our current research projects. Very soon, I plan to start work on Good quality Tamil to Kannada translation and also, Language independent speech recognition, suitable for Indian multilingual scenario.
What do you really want to achieve in Indic Computing?
All Indian people, who do not know English at all, have as much access to the computers and information technology, as others who are conversant with English. Reading disability of any kind (visually challenged or lack of knowledge of a script) is not an impediment for people to read any printed book in any Indian language. Classical literature, music, science and technology and every other information is easily available in Indian languages to any Indian for accelerated and convenient learning. This shall definitely happen in my life time.
What part, do you think, must be played by students and the Indian citizens in general, for rapid development of Indic Computing?
Start learning to type in Indian language keyboards, which are designed to be much more efficient than the current QWERTY keyboard is for English language. Engineering students, especially Computer Science students must create many social projects such as Wikipedia, whereby people with knowledge in any field will be able to create quality content in various fields of knowledge in each Indian language. There must also be such sites, which help experts sitting anywhere in the world to create critical databases such as parallel corpora for machine translation, parts of speech, stress and phrase tagged text corpora for various computational linguistics research studies, voice portals in Indian languages, and exciting educational material in Indian languages involving multimedia.
This site uses Unicode and Open Type fonts for Indic Languages. Powered by Microsoft SharePoint
©2017 Microsoft Corporation. All rights reserved.