Home

Overview

What is Muze?

Muze is an AI-powered music therapy web application that analyzes your emotions in real-time and creates personalized music playlists to match or improve your mood. Using advanced facial recognition technology, Muze detects your emotional state and selects songs from a curated library designed to complement your feelings.

The Problem

Music has a profound impact on our emotional well-being, but finding the right music for your current mood can be challenging. Traditional music apps require manual selection and don't adapt to your emotional state in real-time.

The Solution

Muze combines:

• Computer Vision: Real-time facial emotion detection via webcam
• Machine Learning: Algorithmia's deep learning emotion recognition model
• Music Psychology: Scientifically-organized music library based on emotional clusters
• Adaptive Playlists: Dynamic song selection that responds to mood changes

How It Works

1. Capture: Your webcam captures your facial expression
2. Analyze: AI analyzes your face to detect emotions (Happy, Sad, Angry, Fear, Surprise, Disgust, Neutral)
3. Select: Algorithm selects songs from emotional clusters matching your mood
4. Play: Personalized playlist starts playing automatically
5. Adapt: Every 20 songs, Muze rechecks your emotion and adjusts the playlist

Key Features

🎭 Emotion Detection

• Detects 7 emotions: Happy, Sad, Angry, Fear, Surprise, Disgust, Neutral
• Uses state-of-the-art CNN-based emotion recognition
• Real-time processing via webcam

🎵 Smart Playlist Generation

• 900+ songs organized into 5 emotional clusters
• Cluster-based selection algorithm
• Mood-appropriate song sequences

📊 Emotion Tracking

• Visual emotion history grid
• Color-coded emotion timeline
• Track mood changes over listening sessions

🔄 Continuous Adaptation

• Periodic emotion re-evaluation (every 20 songs)
• Dynamic playlist adjustment
• Responsive to mood shifts

Technology Stack

• Backend: Flask (Python)
• AI/ML: Algorithmia EmotionRecognitionCNNMBP API
• Frontend: HTML5, JavaScript, jQuery
• Media: WebRTC getUserMedia API
• Visualization: Matplotlib
• Image Processing: Pillow (PIL)

Use Cases

• Late Night Listening: Discover music that matches your nocturnal moods
• Emotional Support: Find comfort through mood-appropriate music
• Mood Enhancement: Lift your spirits with energizing tracks
• Stress Relief: Calm down with soothing selections
• Music Discovery: Explore songs you might not have found otherwise

Limitations

• Requires webcam access
• Needs Algorithmia API key (external service)
• Limited to pre-loaded music library (903 songs)
• Emotion detection accuracy depends on lighting and camera quality
• Internet connection required for emotion recognition API

Data Credits

Music dataset sourced from:

• MIREX 2007: Audio Music Mood Classification