A comprehensive, unified dataset for Indian Classical Music research, combining data from multiple sources including Ramanarunachalam, Saraga, and Carnatic Varnam datasets.

RagaSense-Data is the largest unified dataset for Indian Classical Music, containing:

• 1,340+ unique ragas from both Carnatic and Hindustani traditions
• 100,000+ songs with metadata and cross-tradition mappings
• Audio features extracted for machine learning applications
• Cross-tradition mappings validated by musicological experts

• Ramanarunachalam: 868 ragas, 105,339 songs
• Saraga 1.5 Carnatic: 1,982 audio files, 498 metadata files
• Saraga 1.5 Hindustani: Processing in progress
• Saraga Melody Synth: 339 audio files
• Carnatic Varnam: Processing in progress

# Explore the dataset
python3 scripts/exploration/explore_ragasense_data.py

# Web interface
python3 scripts/exploration/web_explorer.py

# Extract audio features
python3 scripts/data_processing/extract_audio_features.py

# Train raga detection model
python3 ml_models/training/gpu_optimized_trainer.py

# Run full validation (CPU, W&B off)
python3 tools/validation/data_validator.py

# Validate a specific tradition
python3 tools/validation/data_validator.py --tradition carnatic

# Validate mappings only
python3 tools/validation/data_validator.py --mappings

# Optional: customize base path and enable GPU/W&B
RAGASENSE_BASE_PATH=$PWD python3 tools/validation/data_validator.py --gpu --wandb

RagaSense-Data/
├── data/                    # Main dataset
│   ├── raw/                # Original data sources
│   ├── processed/          # Cleaned and processed data
│   ├── ml_ready/          # ML training datasets
│   └── exports/           # Community export formats
├── scripts/               # Processing and analysis scripts
│   ├── data_processing/   # Data processing pipelines
│   ├── analysis/          # Data analysis tools
│   ├── exploration/       # Data exploration tools
│   ├── integration/       # Dataset integration
│   └── utilities/         # Utility scripts
├── ml_models/            # Machine learning models
├── tools/                # Development tools
├── docs/                 # Documentation
├── schemas/              # Database schemas
└── website/              # Web interface

Visit our live website: RagaSense-Data

• Dataset Guide
• API Reference
• Contribution Guide

We welcome contributions! Please see our Contribution Guide for details.

This dataset is released under [License Type] for research and educational purposes.

For questions or collaboration, please contact [Contact Information].

• Ramanarunachalam Music Repository
• Saraga Dataset Team
• Carnatic Varnam Dataset Contributors
• Musicological experts who validated cross-tradition mappings