Quark

Important Notes

Before opening the app, make sure you are connected to the internet.
Currently, a quark takes about 1-2 minutes to fully generate your soundtrack — stay patient :)
You can pinch with both hands and stretch to make a quark bigger or smaller to accommodate your physical space
You can also use your hands to clap near a quark to toggle pause
The more colorful your space is — the more colorful your quark will be!
Have fun!

Inspiration

We were fascinated by the idea that every room carries a mood — shaped by lighting, space, texture, and movement. Spatial computing already blends digital and physical worlds, so we wondered: what if your environment could compose its own soundtrack?

What it does

Quark turns your surroundings into a generative soundscape. Walk into a café and hear warm lo-fi tones, step into your kitchen and get airy ambience, or relax in your bedroom with calm pads. The app reads the mood of your environment and composes music and visuals that match it.

How we built it

Passthrough Camera API + OpenAI Vision to interpret room mood and extract semantic cues.
Whisper AI for voice transcription
Audio-visual contextual analysis (K-means + FFT) maps environmental features to color palettes and dynamic parameters that modulate particles and sound
Suno AI for generative music creation, seeded by the environment’s extracted mood vector
OpenAI Speech API for user guidance
Meta XR Audio & Interaction

Challenges we ran into

Latency constraints of generative audio
Balancing the simplicity of a clean and minimal UX with a system with complexity and many moving parts
Tuning the particle system so that it feels alive without overwhelm
Going through several iterations of CPU/GPU optimization