Skip to content

gyoomei/mimoscore

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 

Repository files navigation

🎵 MiMoScore

AI Music Taste Profiler — Powered by Xiaomi MiMo V2.5

Paste songs or add artists. Discover your listener personality. Get AI-powered recommendations based on your unique music DNA.

Live Demo Try Now AI Data License


📖 The Problem

Spotify Wrapped tells you what you listened to for one month. It doesn't tell you who you are as a listener. Why do you gravitate toward melancholic indie at 2am? Why does your playlist mix classical with hip-hop? What does your music taste reveal about your personality?

MiMoScore fixes that. Paste your songs. MiMo V2.5 analyzes the patterns, assigns you a listener archetype, maps your genre DNA, and recommends songs that match your unique taste fingerprint.


✨ How it works

You paste:    "Bohemian Rhapsody - Queen"
              "Blinding Lights - The Weeknd"
              "Lose Yourself - Eminem"
              "Clair de Lune - Debussy"

MiMo writes:  🎭 THE TIME TRAVELER
              You span four decades of music without allegiance
              to any era. Your taste values emotional intensity
              over genre rules — from operatic rock to orchestral
              minimalism. You don't follow trends; you follow feelings.

              Genre DNA: Rock 30% · Pop 25% · Classical 20%
                         Hip-Hop 15% · Electronic 10%

              ✨ Recommended: "Nuvole Bianche" — Ludovico Einaudi
              "Because you crave the same emotional gravity as
              Clair de Lune but with modern cinematic weight."

That's the entire UX — paste, profile, discover, explore.


🎯 Core Features

Capability Detail
🎭 Listener Personality Creative archetype: The Melancholic Dreamer, The Genre Alchemist, The Time Traveler, The Energy Seeker, The Deep Listener...
🧬 Genre DNA Percentage breakdown of your top 5-6 genres with visual bars
🌈 Mood Radar 8-dimension mood chart (energetic, calm, dark, uplifting, melancholic, romantic, angry, dreamy) on canvas
📊 Music Stats Avg BPM, energy, danceability, positivity, acousticness
🎤 Artist Search Search any artist via Deezer API — auto-fetches top tracks
🎶 Audio Previews 30-second preview clips from Deezer
AI Recommendations 5 personalized song suggestions with "why" explanations
💬 Discuss with MiMo Chat about your taste, get mood-based suggestions
🌗 Dark/Light Mode Spotify-green accent, WCAG-AA compliant
🌐 Bilingual EN/ID Full Indonesian + English translation
📱 Mobile Responsive Fluid from 375px to 1920px

🧬 The Analysis Pipeline

Step What happens
1. Input Paste songs (one per line) or search artist names
2. Deezer Lookup Each song/artist → real track data (title, duration, rank, preview)
3. Artist Deep Dive Each artist → top 10 tracks fetched automatically
4. MiMo Analysis Track list → personality, genres, moods, stats, decades, recommendations
5. Visualization Render radar chart, genre bars, stats pills, recommendation cards

🏗️ Architecture

┌─────────────────────────────────────────────────────────────┐
│  Input: Song names or artist names                          │
│                          ↓                                   │
│  Deezer Search API        →  track metadata + preview URLs   │
│  (free, no auth, CORS)       artist top tracks               │
│                          ↓                                   │
│  Profile Builder          →  aggregated track summary        │
│                          ↓                                   │
│  MiMo V2.5 (Pollinations.ai) → personality + genres + moods  │
│                               + stats + recommendations      │
│                          ↓                                   │
│  Canvas Renderer          →  mood radar chart (2D)           │
│  CSS Renderer             →  genre bars, stats pills         │
│                          ↓                                   │
│  Chat (Pollinations.ai)   →  ongoing taste discussion        │
└─────────────────────────────────────────────────────────────┘

Zero backend. Everything runs client-side. Deezer API is free with CORS *. No API key. No tracking.


💡 Architecture Decisions

Why Deezer instead of Spotify? Spotify Web API requires OAuth + app registration. Deezer's search and artist endpoints are completely free, no auth, CORS-enabled, and return rich metadata (title, artist, album, duration, rank, preview URL). Perfect for a zero-friction demo.

Why single HTML? Zero deploy friction. One file = one commit = one deploy. Clone and self-host in one minute.

Why listener personality? Numbers (BPM, energy) are boring. Archetypes ("The Time Traveler", "The Melancholic Dreamer") create emotional connection. Users share personality types — they don't share decimal scores.

Why canvas for the radar chart? CSS/SVG radar charts with 8 axes require complex polygon math. Canvas 2D is simpler, more performant, and scales perfectly to any number of mood dimensions.


🧪 Try these examples

Input Expected Personality
Queen, The Weeknd, Eminem, Ed Sheeran, Nirvana The Time Traveler — spans decades without allegiance
Radiohead, Bon Iver, Phoebe Bridgers, Sufjan Stevens The Melancholic Dreamer — emotional depth over everything
Drake, Travis Scott, Bad Bunny, The Weeknd The Energy Seeker — vibes and momentum
Bach, Debussy, Ludovico Einaudi, Max Richter The Deep Listener — values compositional craft
Blackpink, TWICE, BTS, NewJeans The Genre Alchemist — one genre, infinite variations

🛠️ Stack


🚀 Quick Start

git clone https://github.com/gyoomei/mimoscore.git
cd mimoscore
python3 -m http.server 8080
# Open http://localhost:8080

Or just visit the live demo.


📄 License

MIT — know your taste, share your taste.


Built with 🧠 Xiaomi MiMo V2.5 · Music data via Deezer API · Submitted to the Xiaomi MiMo 100T program

About

AI music taste profiler • Paste songs or Spotify playlist • Discover your listener personality • Genre DNA • Mood analysis • AI recommendations • Powered by Xiaomi MiMo V2.5 • Free • Single HTML

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages