An innovative project combining traditional Kurdish music with artificial intelligence to automatically classify Kurdish musical genres with 92% accuracy using Convolutional Neural Networks (CNN) and Deep Neural Networks (DNN).

Project Overview

• Developed a system for classifying 8 distinct Kurdish music genres
• Created a comprehensive dataset of 880 music samples
• Implemented and compared two neural network architectures (CNN & DNN)
• Achieved 92% classification accuracy using CNN
• Built a web application for real-time music genre classification

Technical Highlights

• Converted audio data to mel-spectrograms for visual processing
• Utilized K-Fold Cross-Validation for model optimization
• Processed 30-second samples divided into one-second intervals
• Implemented data augmentation techniques for better model training
• Developed using Python, TensorFlow, and web technologies such as React.js

Impact

This project represents the first systematic attempt to classify Kurdish music genres using AI, contributing to both Kurdish cultural preservation and modern music information retrieval technology.

Try It Out: Experience the classifier yourself at Kurdish Music Classifier (KMC)

Learn More:

• Read the full research paper published in Engineering Proceedings of MDPI
• Discover the story behind this research in my detailed blog post, where I explain the process step by step.