About Sylvia AI
A closer look at the emotion-adaptive AI companion that powers the Artificial Human experience.
Sylvia AI began as a personal challenge — to build an assistant that doesn’t just listen, but truly reacts to how someone feels. I grew up trying to understand the quieter emotions in people around me, and that curiosity slowly grew into a technical question: Can a machine sense mood and adapt its tone the way humans do?
This project is my attempt to explore that question. It became a blend of expression recognition, responsive dialogue, and personalized music generation — all guided by a simple idea: technology should meet people where they are emotionally. Sylvia is my way of turning the quiet emotions I struggled to decode as a kid into something that can help others feel understood.
1. Mood Detection
Using real-time facial analysis via face-api.js, Sylvia identifies expressions such as happiness, sadness, fear, surprise, or neutrality.
The system chooses the dominant expression with a confidence score and lets the user confirm or retry.
2. Adaptive Conversation
Once the mood is locked, Sylvia adjusts her conversational tone accordingly — warm, energetic, comforting, calm, or reflective —
creating dialogue that feels emotionally aware rather than generic.
3. Music Intelligence
Sylvia offers four mood-driven music pathways: therapeutic transitions, nostalgic soundtracks, weather-based playlists, and custom “Be Your DJ” mixes.
These are powered through curated logic and playlist-generation APIs.
4. Frontend + Deployment
The interface is built with HTML, CSS, and JavaScript, with emotion scanning rendered live through the camera stream.
The entire project runs on cloud deployment through Render.
I designed and developed the full system — from mood detection logic to Sylvia’s conversational flow and the music engine. All UI elements, animations, visual styles, transitions, and page architecture were built manually without templates.
Every part of Sylvia — her voice, her reactivity, the experience around her — is the result of many iterations, experiments, late-night debugging sessions, and a lot of coffee.
In the first 100 people who used Artificial Human, something surprising happened. They didn’t treat it like a demo — they stayed, they engaged deeply, and several even returned the next day. For an early-stage prototype, seeing real users connect with it this quickly felt like the first signal that the idea had real weight.
What stood out the most wasn’t the analytics spike, but the messages people sent afterwards. Users said Sylvia felt “surprisingly grounding after a long day.” Some returned to use it as a space to vent emotions they had no one else to tell. In the first 36 hours alone, the platform recorded over 600 interactions, with users spending real time in conversation instead of bouncing away in seconds.
Even though every session resets, the calm people found in those moments didn’t. A few users shared that Sylvia helped them feel less alone, sleep better, or simply breathe for a second in a chaotic routine.
For something that started as a quiet personal curiosity, Artificial Human has already become a small emotional buffer for the community around it. This early impact is what keeps me building — not the numbers, but the fact that even a short conversation can make someone’s evening a little lighter.
This is still just the beginning — a prototype people unexpectedly found comfort in. And that early human response is what I’m most proud of.
I plan to expand Sylvia with voice-based emotion sensing, better memory, deeper personalization, and a more nuanced emotional model that can react to subtle changes in tone and facial micro-expressions.
Sylvia is still early — but she’s a step toward the kind of emotionally-aware AI I hope to build in the future.
Sylvia never stores photos, camera input, or conversation data.
All emotion detection runs locally on the user’s device, and chat sessions reset automatically after each use.
No personal information, facial data, or messages are saved on the server, and nothing is shared with third parties.
The goal is simple: offer emotional support without collecting anything in return — using the platform is fully anonymous.