DATASET_INFO.md

# Music Classification Dataset - Information



## 📋 Dataset Overview

This comprehensive music classification dataset contains audio samples from 8 different music genres, designed for machine learning and audio analysis tasks.

### Dataset Specifications

| Attribute | Value |
|-----------|-------|
| **Total Samples** | 1000+ audio files |
| **Genres** | 8 (Classical, Jazz, Rock, Pop, Hip-Hop, Electronic, Country, Blues) |
| **Sample Duration** | 30 seconds |
| **Sample Rate** | 22050 Hz |
| **Audio Format** | WAV (lossless) |
| **Bit Depth** | 16-bit |
| **Channels** | Mono |
| **Split Ratio** | 80% Training, 20% Testing |

## 🎵 Genre Information

### 1. Classical
- **Sample Count**: 125+ tracks
- **Characteristics**: Orchestral compositions, chamber music, symphonies
- **Tempo Range**: 60-140 BPM
- **Examples**: Bach, Mozart, Beethoven styles

### 2. Jazz
- **Sample Count**: 125+ tracks
- **Characteristics**: Swing rhythms, improvisation, complex harmonies
- **Tempo Range**: 100-180 BPM
- **Sub-genres**: Bebop, Swing, Fusion, Smooth Jazz

### 3. Rock
- **Sample Count**: 125+ tracks
- **Characteristics**: Electric guitars, drums, strong beat
- **Tempo Range**: 110-160 BPM
- **Sub-genres**: Classic Rock, Hard Rock, Alternative

### 4. Pop
- **Sample Count**: 125+ tracks
- **Characteristics**: Catchy melodies, mainstream appeal
- **Tempo Range**: 100-140 BPM
- **Sub-genres**: Dance Pop, Contemporary Pop, Synth Pop

### 5. Hip-Hop
- **Sample Count**: 125+ tracks
- **Characteristics**: Rap vocals, sampling, strong bass
- **Tempo Range**: 80-110 BPM
- **Sub-genres**: Rap, Trap, Old School, Underground

### 6. Electronic
- **Sample Count**: 125+ tracks
- **Characteristics**: Synthesizers, electronic beats, digital production
- **Tempo Range**: 120-150 BPM
- **Sub-genres**: House, Techno, Trance, Ambient

### 7. Country
- **Sample Count**: 125+ tracks
- **Characteristics**: Acoustic guitars, storytelling lyrics
- **Tempo Range**: 90-130 BPM
- **Sub-genres**: Traditional Country, Modern Country, Bluegrass

### 8. Blues
- **Sample Count**: 125+ tracks
- **Characteristics**: 12-bar blues, guitar solos, emotional expression
- **Tempo Range**: 60-120 BPM
- **Sub-genres**: Delta Blues, Electric Blues, Chicago Blues

## 📊 Extracted Features

### Temporal Features
- **MFCC** (Mel-Frequency Cepstral Coefficients): 13 coefficients
- **Zero Crossing Rate**: Rate of sign changes in signal
- **RMS Energy**: Root mean square energy
- **Tempo**: Beats per minute (BPM)

### Spectral Features
- **Spectral Centroid**: Center of mass of spectrum
- **Spectral Rolloff**: Frequency below which 85% of energy lies
- **Spectral Bandwidth**: Width of the frequency band
- **Spectral Contrast**: Difference between peaks and valleys
- **Spectral Flatness**: Tonality coefficient

### Harmonic Features
- **Chroma Features**: 12-dimensional pitch class profile
- **Harmonic-to-Noise Ratio**: Measure of periodicity
- **Tonnetz**: Tonal centroid features

## 📂 File Structure

```
music-classification/
├── data/
│ ├── audio/ # Audio files organized by genre
│ │ ├── classical/
│ │ │ ├── classical_001.wav
│ │ │ ├── classical_002.wav
│ │ │ └── ...
│ │ ├── jazz/
│ │ ├── rock/
│ │ ├── pop/
│ │ ├── hiphop/
│ │ ├── electronic/
│ │ ├── country/
│ │ └── blues/
│ ├── train_data.csv # Training set metadata
│ ├── test_data.csv # Test set metadata
│ └── features.csv # Pre-extracted features
├── models/
│ ├── train_model.py # Model training script
│ ├── predict.py # Prediction script
│ └── saved_models/ # Trained model files
├── utils/
│ ├── audio_processor.py # Audio processing utilities
│ └── feature_extractor.py # Feature extraction functions
├── notebooks/
│ ├── exploratory_analysis.ipynb # Data exploration
│ └── audio_visualization.ipynb # Audio visualizations
├── requirements.txt # Python dependencies
└── README.md # Documentation
```

## 🔧 Usage Examples

### Loading the Dataset

```python
import pandas as pd

# Load training data
train_df = pd.read_csv('data/train_data.csv')
print(f"Training samples: {len(train_df)}")
print(f"Genres: {train_df['genre'].unique()}")

# Load features
features_df = pd.read_csv('data/features.csv')
print(f"Features shape: {features_df.shape}")
```

### Processing Audio Files

```python
from utils.audio_processor import AudioProcessor
from utils.feature_extractor import FeatureExtractor

# Initialize processors
audio_proc = AudioProcessor()
feature_ext = FeatureExtractor()

# Load and process audio
audio = audio_proc.load_audio('data/audio/jazz/jazz_001.wav')
features = feature_ext.extract_all_features(audio)

print(f"MFCC: {features['mfcc'].shape}")
print(f"Tempo: {features['tempo']} BPM")
```

### Training a Model

```bash
# Train Random Forest model
python models/train_model.py --model random_forest

# Train SVM model
python models/train_model.py --model svm

# Train KNN model
python models/train_model.py --model knn
```

### Making Predictions

```bash
# Predict genre for an audio file
python models/predict.py --audio sample_music.wav --model random_forest
```

## 📈 Performance Benchmarks

### Expected Model Performance

| Model | Accuracy | Precision | Recall | F1-Score |
|-------|----------|-----------|--------|----------|
| Random Forest | 85-90% | 0.87 | 0.85 | 0.86 |
| SVM | 82-87% | 0.84 | 0.82 | 0.83 |
| KNN | 78-83% | 0.80 | 0.78 | 0.79 |
| Neural Network | 88-93% | 0.90 | 0.88 | 0.89 |

## 🎯 Applications

1. **Music Streaming Services**
- Automatic playlist generation
- Music recommendation systems
- Content categorization

2. **Music Production**
- Genre-based music generation
- Style transfer
- Audio effects optimization

3. **Research & Education**
- Music information retrieval studies
- Audio signal processing courses
- Machine learning tutorials

4. **Content Management**
- Automatic music library organization
- Copyright detection
- Content moderation

## 📚 References & Resources

### Academic Papers
- "Musical Genre Classification of Audio Signals" - Tzanetakis & Cook (2002)
- "Music Genre Classification: A Comparative Study" - Sturm (2014)
- "Deep Learning for Audio Signal Processing" - Purwins et al. (2019)

### Useful Links
- [Librosa Documentation](https://librosa.org/)
- [Music Information Retrieval](https://musicinformationretrieval.com/)
- [Audio Content Analysis](https://www.audiocontentanalysis.org/)

## 👥 Author Information

**Name**: Molla Samser
**Company**: RSK World
**Role**: Founder & Developer
**Designer & Tester**: Rima Khatun

### Contact Details
- **Website**: [https://rskworld.in](https://rskworld.in)
- **Email**: help@rskworld.in, support@rskworld.in
- **Phone**: +91 93305 39277

## 📄 License & Usage Terms

This dataset is provided for **educational purposes only**.

### Terms of Use:
- ✅ Educational and research use
- ✅ Non-commercial projects
- ✅ Academic publications (with citation)
- ❌ Commercial use without permission
- ❌ Redistribution without attribution

### Citation

If you use this dataset in your research or project, please cite:

```
@dataset{music_classification_2026,
title={Music Classification Dataset},
author={Molla Samser},
organization={RSK World},
year={2026},
url={https://rskworld.in}
}
```

## ⚠️ Disclaimer

Content used for educational purposes only. For full disclaimer, visit:
[https://rskworld.in/disclaimer.php](https://rskworld.in/disclaimer.php)

---

**© 2026 RSK World - Free Programming Resources & Source Code**

*Founded by Molla Samser with Designer & Tester Rima Khatun*

---

For questions, support, or contributions, please contact us at help@rskworld.in

Theme Settings

Color Scheme

Display Options

Font Size

DATASET_INFO.md