A beautiful, minimalist AI-powered dashboard that reveals deep insights into your Spotify listening habits using AI-powered pattern detection and statistical analysis.
- Audio Feature Analysis: Energy, valence, danceability, tempo, acousticness with interactive tooltips
- Temporal Patterns: When you listen, hourly and daily breakdowns with beautiful charts
- Top Artists & Genres: Your most-played artists and favorite genres with album art
- Recently Played Timeline: Track-by-track listening history with timestamps
- Diversity Metrics: Artist diversity, genre variety, listening velocity
Uses GPT-4o-mini to analyze your listening data and detect 3-5 specific behavioral patterns:
- Temporal Habits: Time-based patterns (morning energy boost, late-night chill)
- Mood Habits: Emotional preferences in music selection
- Genre Habits: Loyalty vs exploration tendencies
- Behavioral Habits: Context clues about how you use music
Each habit includes:
- Catchy, specific title
- Detailed 2-3 sentence description
- Confidence score (0-100%) with visual progress bar
- Color-coded category classification
Creates a rich, personalized listener profile using GPT-4o-mini:
- Archetype: Creative 2-3 word persona (e.g., "Melancholic Explorer", "Energetic Curator")
- Personality: Deep dive into your music personality and psychological profile
- Listening Style: How you use music (background, focus, emotional regulation)
- Personalized Recommendations: 3 specific, actionable suggestions for music discovery
- Custom Cursor Follower: Smooth green glowing dot that follows your cursor
- Interactive Tooltips: Hover over "?" icons to learn what each metric means
- Spotify Design System: Official colors, glass morphism, smooth animations
- Responsive Charts: Chart.js visualizations with radar, bar, and line charts
- Dark Theme: Easy on the eyes with Spotify's signature dark aesthetic
All charts use minimalist Spotify-themed design with different accent colors:
- Top Genres - Bar chart (Spotify Green)
- Listening by Hour - Line chart (Accent Blue)
- Audio Features - Radar chart (Accent Purple)
- Mood Distribution - Doughnut chart (Multi-color)
- Top Artists - List with popularity scores
- Node.js + Express: API server with Spotify OAuth
- Python: Statistical analysis engine
- NumPy: Numerical computations
- Pandas: Time-series analysis
- Statistical calculations
- OpenAI GPT-4o-mini: AI pattern detection
- Vanilla JavaScript: No frameworks, pure performance
- Chart.js: Beautiful, responsive charts
- Spotify Design System: Official colors and styling
Spotify API → Express Backend → Python Analytics → AI Analysis → Chart.js Visualizations
- Node.js v14+
- Python 3.11+
- OpenAI API key
- Spotify Developer account
-
Ensure all packages are installed (already done):
npm install chart.js plotly.js-dist d3 pip install numpy pandas matplotlib seaborn plotly scikit-learn scipy
-
Configure Spotify App (if not already done):
- Go to https://developer.spotify.com/dashboard
- Ensure Redirect URI includes:
http://127.0.0.1:3000/callback
-
Environment Variables (already configured in
.env):OPENAI_API_KEY=your-key SPOTIFY_CLIENT_ID=your-id SPOTIFY_CLIENT_SECRET=your-secret SPOTIFY_REDIRECT_URI=http://127.0.0.1:3000/callback
-
Start the server:
node server.js
You should see:
🎧 Spotify Analytics Server running on http://localhost:3000 -
Open the dashboard:
- Navigate to:
http://127.0.0.1:3000/index.html
- Navigate to:
-
Sign in with Spotify:
- Click "Sign in with Spotify"
- Authorize the app
- Explore your music analytics!
Note: The app will show an account selection dialog every time you sign in, allowing you to easily switch between Spotify accounts.
- User authenticates via Spotify OAuth
- Backend fetches:
- Last 50 recently played tracks
- Top 50 tracks (medium-term)
- Top 50 artists (medium-term)
- Audio features for all unique tracks
The app implements an intelligent caching system to preserve AI-generated habits and persona data:
- Data Hash Generation: When analytics data is fetched, a SHA-256 hash is created from key metrics (top artists, genres, audio profile, etc.)
- Cache Check: Before calling OpenAI API, the system checks if:
- Cached insights exist for this user (keyed by Spotify user ID)
- The data hash matches the current data
- Smart Regeneration: AI insights are only regenerated when:
- User's Spotify data has changed (different top artists, genres, etc.)
- First time a user accesses the dashboard
- Cache has been cleared
- Performance Benefits:
- Instant loading of AI insights on page refresh
- Persists across sign-out/sign-in - returning users get cached insights immediately
- Reduced OpenAI API costs
- Consistent user experience
- Only updates when meaningful data changes occur
The cache is stored in-memory per Spotify user ID (not per session) and includes:
dataHash: SHA-256 hash of analytics datahabits: Cached AI-generated habitspersona: Cached AI-generated personalastUpdated: Timestamp of last cache update
This means users will see their cached AI insights even after signing out and back in, as long as their listening data hasn't changed significantly.
Computes:
- Statistical summaries: Mean, median, std dev for audio features
- Top lists: Most-played artists and genres
- Temporal analysis: Hourly/daily patterns, energy by time
- Mood classification: % of tracks that are energetic, calm, happy, sad, danceable
- Diversity score: Herfindahl index of genre variety
Prompt engineering pattern that:
- Receives statistical data
- Analyzes for behavioral patterns
- Returns 3-5 specific habits with confidence scores
- Categorizes each habit (temporal, mood, genre, behavior)
Creative prompt pattern that:
- Synthesizes listening data into personality traits
- Creates a unique archetype name
- Describes listening style and personality
- Generates personalized music recommendations
- Chart.js renders interactive, responsive charts
- Spotify color palette with accent colors for different data types
- Minimalist design with smooth animations
- Mobile-responsive grid layout
Spotify Green: #1DB954 (primary actions, accents)
Spotify Black: #191414 (background)
Dark Gray: #282828 (cards)
Light Gray: #b3b3b3 (text)
Accent Blue: #509BF5 (temporal data)
Accent Purple: #9D4EDD (features, persona)
Accent Orange: #FF6B35 (energy, moods)
Accent Pink: #FF006E (special highlights)Fetches and analyzes user listening data
- Request:
{ sessionId: string } - Response:
{ "analytics": { /* all computed metrics */ }, "top_artists": [ /* artist data with images */ ] }
AI-powered habit detection
- Request:
{ analytics: object, sessionId: string } - Response:
{ "habits": [ { "title": "Morning Energy Boost", "description": "...", "confidence": 85, "category": "temporal" } ] }
AI-generated listener persona
- Request:
{ analytics: object, sessionId: string } - Response:
{ "persona": { "archetype": "Melancholic Explorer", "personality": "...", "listening_style": "...", "recommendations": ["...", "...", "..."] } }
index.html # Minimalist dashboard UI with custom cursor & tooltips
server.js # Express backend with AI endpoints & Spotify OAuth
analyze.py # Python statistical analysis engine (optional)
.env # API credentials (Spotify + OpenAI)
package.json # Node dependencies
README.md # This file
- Calculated using inverted Herfindahl index
- Higher = more varied taste across artists/genres
- Lower = focused on fewer artists/genres
- Energy (0-1): Intensity and activity
- Valence (0-1): Musical positiveness (happy vs sad)
- Danceability (0-1): How suitable for dancing
- Tempo (BPM): Speed of the track
- Acousticness (0-1): Acoustic vs electronic
Percentage of tracks that match each mood:
- Energetic: Energy > 0.6
- Calm: Energy < 0.4
- Happy: Valence > 0.6
- Sad: Valence < 0.4
- Danceable: Danceability > 0.7
This project uses carefully crafted prompt patterns to extract meaningful insights from listening data. Below are the templates and patterns used:
Pattern Type: Structured Analysis with Constrained Output
System Prompt:
You are a music psychology expert analyzing listening patterns. Always respond with valid JSON only.
User Prompt Template:
Analyze this Spotify listening data and identify 3-5 specific behavioral patterns or habits.
For each habit, provide:
- title: A catchy, specific name (e.g., "Morning Energy Boost" not "Likes energetic music")
- description: 2-3 sentences explaining the pattern and what it reveals
- confidence: A score 0-100 indicating how strong this pattern is
- category: One of [temporal, mood, genre, behavior]
Listening Data:
- Audio Profile: Energy X%, Danceability Y%, Positivity Z%, Acousticness W%
- Top Genres: [genres list]
- Artist Diversity: X%
- Time Patterns: Peak hour H, Peak day D
- Tracks per day: X
Respond with ONLY valid JSON in this exact format (no markdown, no code blocks):
{
"habits": [
{
"title": "Habit name",
"description": "Detailed description",
"confidence": 85,
"category": "temporal"
}
]
}
Key Pattern Elements:
- Specificity Instruction: "catchy, specific name" with example to avoid generic outputs
- Structured Output: Clear JSON schema prevents hallucination
- Confidence Scoring: Forces AI to assess pattern strength
- Category Classification: Limits creative freedom to maintain consistency
- No Markdown: Explicitly prevents
jsonwrappers for easier parsing
Pattern Type: Creative Synthesis with Bounded Creativity
System Prompt:
You are a music psychology expert creating listener personas. Always respond with valid JSON only.
User Prompt Template:
Create a detailed music listener persona based on this Spotify data.
Listening Data:
- Audio Profile: Energy X%, Danceability Y%, Positivity Z%, Acousticness W%
- Top Genres: [genres list]
- Top Artists: [artists list]
- Artist Diversity: X%
- Loyalty Score: Y%
- Mainstream Score: Z/100
- Peak listening: Day at Hour:00
Provide:
- archetype: A creative 2-3 word persona name (e.g., "Melancholic Explorer", "Energetic Curator")
- personality: 3-4 sentences describing their music personality
- listening_style: 2-3 sentences about how they use music
- recommendations: Array of 3 specific, actionable music discovery suggestions
Respond with ONLY valid JSON in this exact format (no markdown, no code blocks):
{
"persona": {
"archetype": "Persona Name",
"personality": "Description...",
"listening_style": "Description...",
"recommendations": ["Suggestion 1", "Suggestion 2", "Suggestion 3"]
}
}
Key Pattern Elements:
- Bounded Creativity: "2-3 word" constraint with examples
- Length Control: Sentence count limits prevent rambling
- Specific vs Generic: "specific, actionable" recommendations vs vague suggestions
- Example-Driven: Examples like "Melancholic Explorer" guide tone and style
- Array Structure: Forces exactly 3 recommendations, not more/less
- Role Assignment: "You are a music psychology expert" establishes expertise
- Output Format Enforcement: "Always respond with valid JSON only" prevents format drift
- Few-Shot Learning: Examples show desired output style without full training data
- Constraint-Based Generation: Limits (word counts, categories) prevent hallucination
- Explicit Anti-Patterns: "no markdown, no code blocks" prevents common AI formatting
- Data Contextualization: Providing all metrics gives AI full context for analysis
- Temperature Tuning:
- Habits: 0.7 (balanced creativity/consistency)
- Persona: 0.8 (more creative for unique archetypes)
// Clean markdown code blocks if AI ignores instructions
const responseText = completion.choices[0].message.content.trim();
const jsonText = responseText
.replace(/```json\n?/g, '')
.replace(/```\n?/g, '')
.trim();
const result = JSON.parse(jsonText);This defensive parsing handles cases where the AI adds markdown despite instructions.
- Time-range selector (4 weeks, 6 months, all time)
- Export analytics as PDF/PNG
- Comparison with friends
- Predictive models (next likely song/genre)
- Playlist generation based on insights
- Historical trend tracking
- Advanced visualizations (calendar heatmap, network graphs)
- Sentiment analysis of lyrics
- Concert recommendations based on listening habits
Built with:
- Spotify Web API for music data
- OpenAI GPT-4o-mini for AI insights
- Chart.js for visualizations
- Node.js + Express for backend
- Vanilla JavaScript for frontend (no frameworks!)
Designed with Spotify's official color palette and minimalist aesthetic.
- Audio Features 403 Error: If your Spotify app is in Development mode, audio features may return 403. The app gracefully handles this by using sensible defaults.
- Account Switching: The
show_dialog: trueparameter forces Spotify to show the authorization dialog every time, making it easy to switch accounts. - Development Mode: Remember to add your Spotify account to the app's allowed users in the Spotify Developer Dashboard.
Enjoy discovering your Music DNA! 🎵