Automatic Classification of Bird Calls

Project under Dr. Sarang C Dhongdi Jan 2019 – May 2019

• Studied algorithms that use signal processing techniques to identify bird species from their calls or songs, and learnt about transforms like the mel frequency cepstral coefficients.
• Implemented machine learning algorithms for the same using python. Learnt about convolutional neural networks and compared results with traditional algorithms using only signal processing methods.

Voice Scrambler and Unscrambler

As Part of ’Digital Signal Processing’ Course Aug 2018 – Dec 2018

• Designed a system that scrambles and unscrambles voice signal for safe, loss­less transmission. The process primarily involved modulation with a key frequency, and included 2 low pass filters.
• The model was first tested on MATLAB and finally implemented on TI 6748 DSP kit. It was tested on recorded as well as real­time audio data.
• Correlation between original and scrambled signal was found to be 0.05, while correlation between original and unscrambled signal was found to be 0.98.
• Checkout the project report here

Spherical Harmonics and its Applications

Project under Dr. Pradeep Boggarapu Jan 2019 – May 2019

• Mathematically studied spherical harmonics and their application in signal processing, specifically, head-related transfer functions (HRTF).
• Studied one paper that analyses the effect spatial smoothing has on the localization accuracy. It was done by systematically reducing the order of a spherical-harmonic-based HRTF representation.
• The results suggested that listeners do not rely on the fine details in an HRTF’s spatial structure and imply that some of the theoretically-derived bounds for HRTF sampling may be exceeding perceptual requirements.
• Checkout the project report here

Language Generator

As part of audited Machine Learning course

• Used Markov Chains to predict the next character in a paragraph based on the previous 40 characters.
• Checkout project on github here