Music Genre Classification

This project includes classifiers for music genres. The dataset is organized into the dataset directory, which contains ten subdirectories with different music genre, each with 50 WAV files. The project utilizes three different classification models and employs five feature extraction methods.

Audio Features Used

1. Mel-Spectrogram
2. Mel-Frequency Cepstral Coefficients (MFCC)
3. Spectral Contrast
4. Zero Crossing Rate
5. Chroma STFT

Classification Models

1. K-Nearest Neighbors (KNN)
2. Support Vector Machine (SVM)
3. Random Forest

Usage

1. Clone or download this project.
2. Install the required dependencies:

   pip install -r requirements.txt
   

3. Run the main script:

   python main.py
   

4. Customize the classifier and feature extraction method to use in the 'Start Training' section.

Results and Analysis

	KNN	SVM	Random Forest
Mel-Spectrogram	0.296	0.337	0.366
MFCC	0.396	0.562	0.534
Spectral Contrast	0.352	0.43	0.434
Zero Crossing Rate	0.216	0.238	0.216
Chroma STFT	0.238	0.272	0.234

The results show that the MFCC feature achieves the highest accuracy across all classification models.