Sep 2024 – Apr 2025
AI + Music / Mobile App / Thesis Project
I listen to music all the time — on the train, while working, before I fall asleep. It shapes my mood, keeps me company, and makes ordinary moments feel like something more. And I'm not alone. Streaming made music effortless for over 616 million people worldwide. It's been the way we experience music for nearly 20 years.
Subscriber growth has dropped from 29% to just 6%. The old playbook — low prices, same experience for everyone — is running out of steam. At the same time, AI and emotion recognition are advancing fast, but current recommendations still just look at what you listened to last week. They don't know how you feel right now. That gap felt like a design opportunity to me.
Do people feel something is missing in their music experience? Would they even be open to AI playing a role — or would it feel intrusive? Instead of assuming, I talked to real listeners about their daily habits, what they love about current platforms, and what they imagine music could feel like in the future.
I tried building personas and journey maps, but it didn't work — everyone listened so differently that there was no "typical user." And nobody was frustrated with their current experience. They just had this quiet hope that music could feel more personal. So I wasn't fixing a problem — I was designing toward something people hadn't fully imagined yet.
People are open to AI — but only if it's subtle and emotionally resonant, not flashy. And music choices are deeply contextual: time, place, activity, and mood shape what someone wants to hear far more than genre or artist loyalty.
If no one's unhappy, polishing the current experience won't make them pay more. The opportunity is in creating something they'd love to have. Personalization needs to be the foundation — adaptive, seamless, emotionally aware. And music should be treated as a tool for everyday well-being, not just entertainment.
That means smarter discovery that reads your mood and balances the familiar with the new — and AI that feels human, learning quietly and never getting in the way.
I explored many directions — biometric detection, voice input, AI-initiated shifts, manual switching. After prototyping and testing, some ideas landed (Vibe Cards, automatic vs. manual modes) and others didn't (chat interfaces felt disconnected, detection was too effortful). The principles that survived: design for the individual, keep AI transparent, make it feel calm, give users control, and earn trust before asking for anything.
.gif)
User Expectation Map
Based on everything I learned, I designed VibeSync — a mobile app that reads your vibe through context like facial expression, location, activity, and weather, and curates music to match your moment.
The app introduces itself, asks for permissions transparently, and starts matching music to your vibe right away.
A short guide walks you through the basics. The Vibe page reads your context and curates music. The Shift button lets you steer your mood, and privacy controls stay one tap away.
Your current moment becomes a living playlist — tracks chosen by mood, activity, and surroundings, not just listening history.
When your mood changes, Shift tells the app in one tap. You can swap Vibe Cards and edit your personal data anytime.
The app notices when your vibe shifts and suggests a new match. You can track mood changes over time and set reminders to check in.
Review what data VibeSync collects and how it's used — anytime, in plain language.
When new data sources are detected, the app asks before using them. You decide what to share.
During this project, I used AI to help summarize interview notes and suggest design directions. It was good at spotting low-level opportunities — small feature improvements, interaction tweaks. But every time I read its output, something felt missing. It couldn't see the bigger picture. It didn't understand that people weren't asking for better playlists — they were hoping music could feel more like them. That high-level sense of what people truly expect still has to come from a human designer who's willing to sit with messy, contradictory data and find meaning in it.
This project also taught me that AI is changing how people behave — not just how we design. Users are becoming more goal-driven, less linear, and harder to map with traditional UX methods. As designers, we can't just rely on what worked before. But we also can't chase every new technology just because it exists. The real skill is knowing when to slow down, listen to what people actually need, and focus on what genuinely adds value to their lives.