Audio Software Engineering Course Project (May 2022)

Created a real-time C++ JUCE plugin host tailored specifically for multi-band processing, allowing users to effortlessly split stems into frequency bands and apply a diverse range of third-party plugins to each band. Designed a low-complexity multi-band filter with flat group delay response

Audio Software Engineering 

Real-Time AudioFx API in C++ 11, that can be used to apply a variety of effects to a vocal audio signal. The ffects include: CombFilter (FIR, IIR), Vibrato, Tremolo, and Reverb. Reverb implementation is a OLA fast convolution that achieved an average run-time of 5.6s on a 3-minute track with an Impulse Response of 3s

Brain Music Interfaces Course Project (Dec 2021)

Designed and developed a drum machine interface controlled by heart rate and muscle activity. The tempo is set by the user tapping their finger; the user’s real-time changes in heart rate adjust the micro-timing of individual components in the drum groove.

Audio Content Analysis Class Project (Dec 2021) 

Created an educational web application that takes input audio signals containing melodies and generates visually intuitive graph representations, revealing the note-to-note relationships and structural elements in a concise and insightful manner.

Under the guidance of Dr. Ranjani H.G.

In this work we develop an automated learning application based on feedback from a user to promote the learning of basic rhythmic patterns. The application is built to mimic a teacher-student relationship where, the subsequent levels of pattern complexity are generated based on the performance measured on the current pattern played by the user.  The web application built for the project serves as a data collection tool to test my rhythm education hypothesis, kindly do register and help me in my research.

Under the guidance of Dr. Ajey S.N.R.

As part of my final year capstone project, I am working on implementing a novel end-to-end deep convolutional neural network architecture operating on multi-channel raw audio data to localize multiple simultaneously active acoustic sources in space. As proposed by Dr. Harshavardhan Sundar.

Audio Sense is an Alexa game skill that makes use of Alexa Conversations. It aims to help and train players to cope with Auditory Hypersensitivity and distractive environmental sounds. The skill sets up a game environment that can span over multiple settings and levels. The goal of the game is for the player to zone out auditory disturbances in a given environment to discern only the necessary and valid information from the same.

A web-based musical instrument with a mind of its own. 

Uses Google's Magenta.js trained machine learning models to generate it's own sequence of melodies and randomise drum grooves.

There's acute shortage of accessible health care in smaller towns and remote villages of Uttarakhand. We developed a system solution to this problem with the following points focussed upon :

• Generation of a cloud based electronic health record system, implementing aadhar authentication.

• Access to medical records and database for medical professionals using Biometric/Facial ID, for emergencies and diagnostic purposes.

• Telemedicine / Call Centre setup for doctors implementing DTH or 2 way communication technology without being dependent on ISP.

This project is a mechanism based on IOT-automation that can detect cracks in structures (Living Spaces and Flyovers). The system gathers information from the magnitude of vibrations produced by the structure using vibration sensors which relay information to a database for further data collection, analysis, and future reference. After the gathered data is analysed, the concerned authorities are notified about the condition of the structure.