"Indian music genre classification using deep learning is like tuning into a new frequency in the world of AI. Just like a DJ blending beats, I used powerful algorithms to distinguish Bollywood pop from classical tunes. With a 76% accuracy, my models hit the right notes, creating a harmonious fusion of technology and traditional melodies. 🎶🎧 #MusicMeetsMachineLearning"
KaggleX Final Project Presentation with Urvi Haval | Kaggle
🎓📊📈
In this final showcase video for the KaggleX POG mentorship program, Urvi Haval presents her project on Indian music genre classification using deep learning. As a third-year student pursuing a B Tech in electronics and communication engineering, Urvi has a keen interest in machine learning, artificial intelligence, and quantum computing. She is also interning at a Quantum startup, focusing on implementing a Quantum decision tree and researching Quantum machine learning models. With a career aspiration to pursue higher education in artificial intelligence and contribute to research using new technologies, Urvi’s project delves into the classification of sounds in Indian music, exploring methods and libraries for audio processing and feature extraction.
Overview
As part of the KaggleX POG mentorship program, Urvi Haval’s final project focuses on Indian music genre classification using deep learning. The project makes use of lebrusa for audio processing and feature extraction, as well as artificial neural networks and convolutional neural networks for classification.
Background and Inspirations
Urvi has been learning classical music for the past 10 years, thus leading her to find audio processing intriguing, especially within the context of Indian music. Upon observing the lack of projects and YouTube videos based on the classification of sounds in Indian music, Urvi decided to explore different methods and libraries for audio processing, culminating in the development of her project notebook.
Audio Processing and Neural Networks
Throughout the project, Urvi delved into various aspects of audio processing, learning about audio file parameters, basic structure, waveforms, and concepts like Fast Fourier Transform (FFT) and Mel Frequency Cepstral Coefficients (MFCCs). Additionally, she implemented both artificial neural networks and convolutional neural networks to improve the accuracy of the classification model and solve the overfitting problem.
Implementation
Urvi’s project notebook provides a detailed insight into her implementation process, starting with the importation of libraries and loading of sample signals. She proceeded to visualize audio file waveforms, apply Fourier transformation, stft, and MFCCs, leading to the extraction of features for classification using convolutional neural networks. The notebook also highlights the use of kernel regularization to address overfitting issues in the model.
Conclusion
Urvi successfully constructed a project aimed at addressing the classification of Indian music genres through deep learning, achieving an accuracy of 76%. Her thorough exploration of audio processing, neural networks, and model improvement techniques showcases her dedication and proficiency in the field.
Key Takeaways
- The project focused on Indian music genre classification using deep learning.
- Audio processing and feature extraction were carried out using lebrusa.
- Artificial neural networks and convolutional neural networks were implemented for classification and model improvement.
- A detailed notebook was created to document the implementation process and analysis.
FAQ
Q: What was the main goal of the project?
A: The main goal was to classify Indian music genres using deep learning techniques, achieving an accuracy of 76%.
This presentation by Urvi Haval provides a comprehensive overview of her project on Indian music genre classification using deep learning. Through her dedication to exploring audio processing and neural network techniques, Urvi has made significant strides in the field, contributing to the advancement of music classification methodologies.
