Music Genre Classification using CNN

This project uses a Convolutional Neural Network (CNN) built with Keras and TensorFlow to classify music into 10 different genres based on their audio features (MFCCs).

Demo Video

Demo.Video.mp4

Features

• Classifies 10 music genres from the GTZAN dataset.
• Extracts Mel-Frequency Cepstral Coefficients (MFCCs) using Librosa.
• Uses a Keras CNN model for classification.
• Provides a command-line script to predict the genre of any audio file.

Project Structure

• data_and_training.ipynb: Jupyter Notebook for data processing, model training, and evaluation.
• predict.py: Command-line script to classify a new audio file.
• visualize.py: Generate a spectrogram for an audio file
• requirements.txt: A list of all required Python packages.

Setup and Installation

1. Clone the repository:

   git clone https://github.com/debug-soham/Music-Genre-Classification.git
   cd music-genre-classification

2. Download the Dataset:

   • Download the GTZAN Genre Collection dataset from here.
   • Unzip it and place the genres_original folder inside a new Data folder in the project root. The final path should be Data/genres_original/. (Note: The Data folder is not included in this repository due to its size.)

3. Create a Virtual Environment:

   # Make sure you have Python 3.11 installed
   py -3.11 -m venv my_env
   .\my_env\Scripts\activate

4. Install Dependencies:

   pip install -r requirements.txt

Usage

1. Train the Model:

   • Open and run all the cells in the 01_data_and_training.ipynb notebook.
   • This will create two files: data.json and music_genre_classifier.keras.

2. Predict a Genre:

   • Use the predict.py script to classify an audio file:

   python predict.py "path/to/your/song.wav"

3. Visualize a Spectrogram (Optional)

   • Use the visualize.py script to generate a spectrogram for any audio file:

   python visualize.py "path/to/your/song.wav"
   

Technologies Used

• Python 3.11
• TensorFlow / Keras
• Librosa
• Scikit-learn
• NumPy
• Matplotlib