ROOMii
Because your feelings deserve to be heard.
Created on 29th December 2025
•
ROOMii
Because your feelings deserve to be heard.
The problem ROOMii solves
In today’s fast-moving digital world, many people work, study, and live in isolated environments — often spending long hours behind screens without emotional connection or support.
While virtual assistants exist, most of them feel robotic, emotionless, and disconnected from the user's mental state. They provide answers — but they don’t feel the user’s mood, stress, or energy.
As a result, people increasingly experience loneliness, lack of motivation, burnout, and emotional fatigue — especially students, remote workers, and individuals who spend a lot of time alone.
ROOMii solves this problem by creating an emotionally intelligent AI companion that doesn’t just process commands — it understands how you feel.
Everyone's individual character is saved and ROOMii is as same as YOU!! And the thing is IT changes according to people and people's Situations and Emotions!!
Challenges we ran into
While building ROOMii, I ran into quite a few unexpected challenges that really pushed me to think differently. Getting real-time emotion detection to work smoothly was tough at first — the camera feed kept lagging because the model was analyzing every frame. I fixed it by caching emotions for a few seconds, which made the performance much faster. Synchronizing AI replies with voice output was another struggle; ROOMii would sometimes interrupt itself or speak over previous messages. I restructured the backend so it waits for the AI’s full response before generating audio and added a stop button to instantly cancel playback if the user speaks again. These small details made ROOMii feel more natural and human.
Another major challenge was making the 3D model expressive and responsive. The model initially faced the wrong direction, and those text-style faces like ^_^ looked odd in 3D. I fixed this by reorienting the model, adding glowing eyes that change color with mood, and introducing breathing and idle animations for realism. I also dealt with voice API limits and local audio clutter — ROOMii now auto-deletes old files and falls back to backup voices if an API fails. Balancing AI logic, live emotion tracking, and smooth visuals wasn’t easy, but seeing ROOMii respond calmly, emotionally, and in real time made it all worth it.
We are still in devlopment phase, so accuracy will be increased soonn!!!
Tracks Applied (1)
Best Innovation
Technologies used