Tata Consultancy Services (TCS) has a Research Fellowship Program wherein the best forty applicants from all over India are annually sponsored for doctoral programs.
Once selected, the successful applicant receives funding from the company for a maximum period of five years, or until the research work undertaken is completed. The company pays a stipend of Rs. 23000 per month for the first two years and 25000 per month, thereafter.
This year, Kuruvachan K. George, Junior Research Fellow at Amrita School of Engineering, Coimbatore, was selected to receive this prestigious fellowship.
He has undertaken his doctoral research as part of a project titled Robust Features for Text/Language Independent Speaker Recognition funded by the Defense Research and Development Organization.
Amrita Vishwa Vidyapeetham will sign an MoU with TCS to enable Kuruvachan to accept this fellowship.
“Mr. Kuruvachan getting the TCS fellowship is a really good achievement, as TCS selects only 40 students from all over the country, including the IITs, etc.” stated Dr. R. Krishnan, Head of Engineering Research at Amrita. Kuruvachan is conducting his research in Amrita’s Machine Intelligence Research Lab. His thesis advisors are Dr. C. Santhosh Kumar, Associate Professor, Department of Electronics and Electrical Engineering and Dr. K. I. Ramachandran, Professor, Department of Mechanical Engineering.
“I consider the fellowship as an outcome of my being registered for a doctoral program in the Department of Electronics and Electrical Engineering, here at Amrita. I am very delighted and consider it an honor to be selected for this prestigious award, in the early days of my PhD. This has increased my self-confidence and has encouraged me to continue with what I am doing now. It was like a reward for the efforts put forth, thus far,” Kuruvachan stated.
Speaker recognition technology is used to identify one who is speaking by analyzing characteristics in the voice. This finds application in phone banking and trading, password resetting, accessing customer care services, credit card activation, transactions and payments.
Text/language independent speaker recognition systems are also perceived to be of great value for enhancing the country’s security needs.
“I am working to enhance the accuracy of the text/language independent speaker recognition systems by identifying robust features that have better discriminative ability to classify speakers,” Kuruvachan elaborated, sharing
details of his research.
“We also propose the exploration of noise robust features to enhance performance, when the amount of training data is less, as in several national security related applications. We will also derive robust features that are less sensitive to channel characteristics,” he added.
“We can help deploy a speaker recognition system for security related applications, where impostors talk over a telephone network to mislead/confuse security personnel. This may also cast aspersions on the reliability of the intelligence information, derived from the speaker recognition system,” he further underlined.
November 10, 2012
School of Engineering, Coimbatore