EmotiSound is an AI-powered audio analysis tool that "listens" to raw music files, extracts complex acoustic features, and predicts the emotional DNA of the song. Unlike basic mood classifiers, it generates a multi-dimensional emotional fingerprint and even maps music to synesthesia-inspired colors.

• 🎧 Advanced Audio Feature Extraction using Librosa (Danceability, Energy, Valence, Tempo, Brightness, etc.)
• 🧠 Ensemble AI Model – Voting Classifier with 5 powerful algorithms (Random Forest, Gradient Boosting, SVM, KNN, MLP)
• 🧬 Emotion DNA Radar Chart – Visualizes the unique emotional profile of any song
• 🎨 Synesthesia Color Engine – Converts sound frequencies and energy into vivid colors
• ☁️ Fully Containerized – Runs seamlessly with Docker and Hugging Face Spaces

• Backend: Python, Flask
• Machine Learning: Scikit-learn, NumPy, Pandas
• Audio Processing: Librosa, SoundFile, pydub + FFmpeg
• Frontend: HTML5, CSS3, JavaScript, Chart.js
• Deployment: Docker, Gunicorn, Hugging Face Spaces

1. User uploads an .mp3 or .wav file
2. System extracts low-level and high-level acoustic features using Librosa
3. Features are normalized and passed to the pre-trained Ensemble Voting Classifier
4. Model outputs emotion probabilities across 4 core dimensions:
   • Euphoria ⚡
   • Melancholy 🌧️
   • Serenity ☮️
   • Turbulence 🌪️
5. Results displayed as interactive radar chart + synesthesia color palette

• Python 3.9 or higher
• FFmpeg installed and added to system PATH (download here)

1. Clone the repository

   git clone https://github.com/itslovepatel/emotisound-live.git
   cd emotisound-live

2. Install dependencies

   pip install -r requirements.txt

3. (Optional) Retrain the model

   python train_model.py

4. Run the app

   python app.py

   Then open http://127.0.0.1:5000 in your browser.

emotisound-live/
├── templates/
│   └── index.html              # Frontend with Chart.js radar visualization
├── static/
│   └── (CSS, JS, assets)
├── app.py                      # Main Flask application
├── audio_processor.py          # Feature extraction logic
├── train_model.py              # Model training script
├── emotisound_brain.pkl        # Trained ensemble model
├── requirements.txt            # Python dependencies
├── Dockerfile                  # For Docker deployment
├── spotify_tracks.csv          # Training dataset
└── README.md                   # You are here :)

Contributions are very welcome! Ideas for future features:

• Real-time microphone input
• Spotify/YouTube URL analysis
• Mood-based playlist generator
• Export emotion DNA as NFT metadata

Feel free to fork and open a Pull Request!

This project is licensed under the MIT License – see the LICENSE file for details.

Built with 🎵 passion and ☕ caffeine by Love Patel