Hello everyone,
My name is Vishesh Khosla,I am 2nd year undergraduate student at Netaji Subhas University of Technology and I am currently pursuing Instrumentation and Control Engineering.My area of study revolves around the subjects of Electronics and Computer Science.Apart from that I am very proficient in C++(Data Structures and Algorithms) and Python.I have also been a Teaching Assistant for 4 months at Coding Ninjas(An online platform to learn Coding) for C++ and Data Structures.I am very much interested in the field of A.I(specifically ML,Deep Learning and Computer Vision).I also enjoy doing competitive programming.
I got very much fascinated with the concept of computer vision and have been doing it with the help of OpenCV and I would like to contribute to the organisation.
Machine Learning and Deep Learning Background:
I have been doing machine learning and deep learning for the past 6-7 months in python.I have mastered the following concepts:Linear Regression,Logistic Regression,Decision Trees,Random Forests,Naive Bayes,Support Vector machine,Principle Component Analysis,Natural Language Processing,Simple Neural Networks,CNNs,RNNs.I have made some projects with the help of these which is present in my github account(link is present below).
Experience with DNN(for Audio):
I have studied about Audio Preprocessing(Short Time Fourier Transformation,Spectrogram,MFCCS) with the help of librosa and numpy and preparing a dataset for Music genre classification and then applying neural networks to predict the music genre,i was easily able to grasp these topics due to my good understanding in mathematics.
Vision and Understanding of the Project:
These are the following ideas that can be implemented through:
1.Music Genre Classification
2.Speech Recognition
3.Music Instrument Classification
Since i have a good hold over my English(both Verbal and written),i can contribute by making video tutorials.
Questions:
1.What are some other ideas that need to get implemented in the DNN module for Audio IO?
2.Is applying PCA on an audio file a good idea in order to compress the size of the testing data and increase the efficiency of the algorithm as there are less number of features?
CV: