M. VIJAY KUMAR
Bachelor of Technology (B.Tech), 4th year,
B.Tech, Electronics and Communication Engineering 2009 8.78/10 (CPI)
Indian Institute of Technology Guwahati, Assam, India (expected)
Higher Secondary Education 2005 96.5%
Board of Intermediate Education, Andhra Pradesh.
Secondary Education 2003 91.0%
State Board of Secondary Education, Andhra Pradesh.
GRE â€“ 1460/1600 Verbal â€“ 660/800 Quantitative â€“ 800/800 Analytical Writingâ€“ 5.0/6.0
TOEFL iBT â€“ 111/120.
Programming : C, C++, Assembly (8086/8088)
Software Packages : HTK Toolkit, MATLAB, Visual ADSP++, LT SPICE,
Circuit Maker, VHDL, AVR Studio.
Operating Systems : Windows, Linux.
Research Experience and Major Projects
Exploration of New Feature Extraction Methods for Stress Classification in Speech (Ongoing Bachelorâ€™s Dissertation)
Guide: Dr. S. R. M Prasanna, Associate Professor, IIT Guwahati.
The variability introduced by a speaker under stress causes speech recognition systems trained with neutral speech to fail. One of the ways to improve the performance of a speech recognition system under stressful conditions is to first classify the stress and then use appropriate compensation techniques. The success of a stress classification system largely depends upon the set of features used to characterize stress specific information. Cepstral and Linear Prediction (LP) based features which are considered to be highly robust for speech recognition have not been able to classify stress satisfactorily over a wide range of domains. Our aim is to explore new features which can significantly improve the performance of a stress classification system. In the first phase of the project, we have successfully implemented and analyzed the performance of conventional MFCC features and intend to compare it against the sinusoidal based and other such features.
Automatic Speech Recognition (ASR) in a Closed Noisy Environment (Inside Vehicles)
Guide: Dr. Andre Goalic, Assistant Professor, TELECOM- Bretagne, France.
In conventional speech recognition systems, MFCC coefficients concatenated with delta and delta-delta coefficients are normally used. Furthermore, the concatenation of AM and FM features into acoustic features has proved the effectiveness in improving the performance of ASR in noise. Our aim was to investigate the effect of removing the rapidly varying AM and FM components, on an ASR system. By applying Frequency Amplitude Modulation Encoding (FAME) analysis (which has been invoked in cochlear implants) before the feature extraction stage, we could keep only the slowly varying components of AM and FM features and suppress the fast varying ones. A detailed analysis of the ASR results using FAME synthesized speech was done and a comparison against ASR using speech having both slowly and rapidly varying AM and FM components was made. The HTK toolkit and TIDIGITS database were used for implementing the ASR task.
Design of an Automatic Switch for AC/DC Adapter to Turn Off with Equipment
Guide: Dr. H. B Nemade, Associate Professor, IIT Guwahati.
Speech Enhancement using Maximum a Priori (MAP) based Statistical Techniques
(Summer Internship 2007)
Guide: Dr. R. Sinha, Assistant Professor, IIT Guwahati.
The basic aim of the project was to reduce the additive noise without affecting the intelligibility and quality of the speech signal. For optimal performance, generalized gamma modeling of speech based on the IEEE paper: Tran Huy Dat, Kazuya Takeda, and Fumitada Itakura, â€œGeneralized Gamma Modeling of Speech and Its Online Parameter Estimation for Speech Enhancement,â€ ICASSP 2005, was used. Along with the generalized gamma modeling, other conventional MAP based speech enhancement techniques like the Gaussian modeling of speech(on both magnitude square and arbitrary magnitude power domains) were successfully implemented in MATLAB and their performance against a variety of noisy speech signals was compared.
Made a fully autonomous robot capable of navigation in an arena avoiding walls and obstacles (stationary and moving) as a part of Techniche 2007, the annual Technical Festival of IIT Guwahati. Also improved and modified the robot for more efficient navigation and line following as a part of Techniche 2008. Later modified it for shortest path problem, implementing the flood-fill algorithm, in the event Micro-mouse.
Short Term Projects and Papers
Simulation and Hardware Implementation of Digital and Analog Communication Schemes
Performed MATLAB simulation of modulator, demodulator and error analysis for Conventional AM, Double-Sideband Suppressed Carrier AM, Single Sideband AM, FM, M-ary Pulse Amplitude Modulation, M-ary Phase Shift Keying, M-ary Frequency Shift Keying; Implementation of Digital transmitter and receiver for band limited channels, source and channel encoders and decoders.
Also designed a circuit for implementation of Amplitude Modulation/ Demodulation and Frequency Modulation / Demodulation, and did hardware implementation of BPSK and BFSK on the trainer board.
Software projects in Digital Signal Processing
Implementation of equiripple linear-phase FIR filters design, Parks-McClellan filter design, least-mean-square error filter design, and IIR filters design using impulse invariance, bi-linear transformation and frequency transformations.
Electronic Measurement and Instrumentation
Development of circuits for signal conditioning, signal recovery, telemetry, pc-based instrumentation, and computer controlled test systems.
Term papers and Technical Reports
Developed a text editor similar to a notepad which supports basic file opening, editing and saving functions. The source code was written using Data Structures in C.
Cong-Thanh Do, M.Vijay Kumar, Dominique Pastor, Andre Goalic, â€Automatic Speech Recognition of Cochlear Implant-like Spectrally Reduced Speech,â€ Proceedings of the 15th National Conference on Communications (NCC-2009), Guwahati. (Under Review)
Digital Signal Processing (L)
Communication Systems Engineering
Probability and Random Processes
Principles of Communication(L)
Probability and Random Processes
Information Theory and Coding
Signals and Systems
Radio Frequency Laboratory(**L)
Introduction to Electronics(L)
Introduction to VLSI Design(L)
Advanced Topics in Signal Processing
Advanced Topics in Communication Systems
Electronic Measurement and Instrumentation (L)
Microprocessors and Embedded Systems(L)
Analog Electronic Circuits(L)
Introduction to Electrical Engineering(L)
Legend: (L) indicates the course has a Laboratory accompanying it.
(**L) indicates it was an exclusive laboratory course.
Other Relevant Courses
Real Analysis, ODEâ€™s and Linear Algebra
Complex Variables, PDEâ€™s and Integral Transforms
Introduction to Computing
Object-Oriented Programming and Data Structures
Multivariable calculus, and
Secured a grade of AA (10) in Mathematics in all of the first three semesters at IIT Guwahati.
Secured 11th rank in the National Science Talent Search Examination 2002-2003 held by Dr.A.S.Rao Awards Council.
Secured 12th rank in Young Scientist Contest held in 2001-2002.
Organizer of Pradarshini, an event in the exhibition module of Techniche â€™06, the Annual Technical Festival of IIT Guwahati
Member of the Publicity Team of Techniche â€™06, the annual Technical Festival of IIT Guwahati
Participated in Manual Robotics in Techniche â€™05, the annual Technical Festival of IIT Guwahati
Member of the organizing team of Shades, the fine arts event in Alcheringa â€™06, the annual cultural festival of IIT Guwahati.
Selected for Table Tennis by National Sports Organization (NSO).