🎵 Music Genre Classification

Note : This repository is one of the challenges made for the event "ForkThis" organized by the chapter Computer Society of India(CSI), VIT Vellore in the year 2025.

📌 Overview

This project focuses on automatic music genre classification using pre-extracted audio features. The main goal is to train machine learning models that can predict the genre of a track based on its numerical audio characteristics.

📂 Dataset Description

The dataset consists of two main CSV files:

tracks.csv – metadata for tracks. This file contains hierarchical column headers (two rows of headers). From this, only the track ID and the top-level genre label were extracted for use.

features.csv – numerical audio features for each track, with track IDs as indices.

A subset of the dataset was used, focusing only on tracks with a clearly defined top genre.

Dataset link : https://drive.google.com/drive/folders/1tnlfqa4KoZFVeQ-xMq9rHM3i6TJeRFqI?usp=sharing

⚙️ Installation

Clone the repository

git clone https://github.com/a-niveditha/MusicGenreClassification.git

Navigate to the project directory

cd MusicGenreClassification

Create a virtual environment (recommended)

python -m venv venv

Activate the virtual environment

On Windows:

#First run:
cd venv/Scripts

#Then:
./Activate.ps1

On macOS and Linux:

#First run:
cd venv/bin

#Then:
source ./activate

Install project dependencies from requirements.txt

pip install -r requirements.txt