Vowel Detection using Digital Signal Processing
Digital Signal Processing, Spring 2019
Implemented Real Time Vowel Classification on Digital Signal Processor. Performed Fast Fourier Transform (FFT) on a moving window and analyzed the frequency components and concluded the vowel in the voice. The algorithm is robust to both male and female voice. The two vowels were mapped to two LEDs present on the DSP.
Video Link- https://youtu.be/ykrvy-2GrEs
Optimal Bidding in Electricity Market
Inter-IIT Techincal Meet, IIT Madras, January 2018
Part of team developing algorithm to optimally bid in the electricity market. The aim is to optimize the cost and the bid quantity, subjected to various constraints. The bidding parameters are predicted using developed models and machine learning algorithms. The optimal bid quantity is balanced between different sources.
Presentation Link- Google Drive Link
Handwritten Language Translator
Self Project, December 2017
Developed a Machine Learning model in scikit-learn to recognize handwritten letters. Implemented Neural Network with one hidden layer and achieved an average accuracy of 94%. Built a GUI in Python to take user input and process letters, converting it into NumPy array. Currently trained and deployed in Hindi and English, easily implementable to other languages.
Video Link- https://youtu.be/ykrvy-2GrEs
SpamSlam: Prediction Market Model to prevent Spam E-mails
InOut-2017, India’s Biggest Community Hackathon, Bangalore, October 2017
Secured first position in Blockchain category, sponsored by Gnosis. Developed a model which uses the concepts of predictive market to decrease spam by selectively monetizing emails. The idea was implemented by using Gnosis API and Django. Also, an AI agent was built to serve as the referee in this game of prediction market.
Presentation Link- Google Drive Link
Universal Sensor Signal Conditioning
Analog Lab, Spring 2018
Built a circuit to condition a signal into ADC readable format irrespective of source, input range and bandwidth. Designed a multipurpose and re-configurable circuit compatible with DSPs, FPGA and micro-controllers. Also built an additional 4th order Butterworth Filter to prevent aliasing whose cutoff frequency can be changed using switches.
Presentation Link- Google Drive Link
StarBox: Arduino Controlled 8x8x8 LED Cube
Institute Technical Summer Project, Summer 2017
Developed an 8x8x8 LED Cube using Shift Registers and MOSFETs. Also, programmed an Arduino to systematically control blinking of LEDs. Displayed letters, patterns and bar and waves of music on the Cube.
Reaction Game using Digital Logic Circuit
Digital Circuits Lab, Spring 2018
Built a game in VHDL on CPLD board that displays reaction time of the player on a LCD screen. Modelled the game as a Finite State Mealy Machine using the concepts of Register Transfer Level (RTL) for the operation of LEDs and push-buttons, and to display the player’s score on an LCD panel.
Two Bit Multiplier using Digital ICs
Introduction to Electronics, Spring 2017
Developed a 4-bit Multiplier using the conventional shift and add algorithm. The algorithm was implemented without the use of any micro controller, thus using only appropriate logic gates and adders. Minimized the wires in the circuit and prepared the final working model on a breadboard.
Working Demo - https://www.youtube.com/watch?v=aQzDZA82Cog