Skip to content

IshankumarP/Aurabeat

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 

Repository files navigation

🎧 AuraBeat: Emotion-Based Music Recommendation System

AuraBeat is a deep learning-based music recommendation system that tailors playlists to your emotional state — all in real-time — using facial expression recognition. It leverages CNN architectures like ResNet50V2 and VGG16 to detect emotions from your face and recommends songs accordingly.


📌 Table of Contents


📖 Overview

“Music is what emotions sound like.”

AuraBeat brings emotional intelligence to music players by recommending songs that resonate with your current mood, not just your listening history.

Whether you're smiling after a good day or looking stressed from a deadline — AuraBeat catches your emotion and queues the right playlist.


🧠 How It Works

  1. Emotion Detection:

    • Uses facial expression recognition via a webcam.
    • Detects emotions like: Happy, Sad, Angry, Neutral, Surprise, etc.
  2. Deep Learning Models:

    • Custom CNN (66% accuracy)
    • ResNet50V2 (68.5% accuracy)
    • VGG16 (65% accuracy)
  3. Music Mapping:

    • Each emotion is mapped to a song mood (from Spotify dataset).
    • Top 5 songs are recommended based on popularity + emotional alignment.
Detected Emotion Song Mood
Happiness Happy
Sadness, Disgust Sad
Anger, Fear Calm
Surprise, Neutral Energetic

🛠 Tech Stack

  • 🧠 TensorFlow / Keras (ResNet50V2, VGG16)
  • 🖼 OpenCV (facial detection)
  • 📊 Pandas, NumPy (data handling)
  • 🎵 Spotify Mood Dataset
  • 📁 FER2013 Emotion Dataset
  • 🧪 Scikit-learn (F1 Score, Confusion Matrix)

🚀 Getting Started

Prerequisites

Ensure you have Python 3.7+ and pip installed. Then install dependencies:

pip install -r requirements.txt

Running the Project

  1. Launch the notebook:
    jupyter notebook Auranewcode.ipynb
  2. Run each cell in sequence to:
    • Load and preprocess data
    • Train or load models
    • Run emotion detection on webcam/image
    • Get top 5 song recommendations

📊 Results

Model Accuracy
ResNet50V2 68.5%
CNN (Custom) 66%
VGG16 65%

Confusion Matrix Sample (ResNet50V2)

  • High precision for Happiness, Surprise
  • Some overlap between Anger & Sadness due to similar facial features

📎 Files

File Description
Auranewcode.ipynb Main Jupyter Notebook with full pipeline
Auranew.pdf Project Report / Technical Documentation
requirements.txt Dependencies list (to be added)
screenshots/ (Optional) Visual outputs or UI demos

📚 References

Includes work inspired by:

  • FER2013 Dataset
  • Spotify Audio Features
  • ResNet / VGG architectures
  • Emotion models (Thayer’s valence-arousal plane)

Full list in Auranew.pdf


📌 Future Work

  • Build a live web app interface (Streamlit or Flask)
  • Integrate Spotify API for real playback
  • Improve accuracy with ensemble models
  • Consider multi-modal emotion input (text, voice, etc.)

🙏 Acknowledgments

Thanks to the CSE(AIML) department for support and feedback throughout the project.


About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors