EchoCalm

EchoCalm — an AI-powered avatar chatbot that helps people struggled with audio hallucination to manage distressing voices and find emotional balance, giving calm to the noise.

Comment

Inspiration

EchoCalm was inspired by Avatar Therapy, where introduced by Prof. Julian Leff at UCL link . This approach creates a digital representation of auditory hallucinations that sufferers hear and interact with them in a safe and guided environment, with aid of human therapist. My motivation is to create a private, accessible and AI-driven alternative for those who face barriers for traditional therapy, due to limitations of cost, social stigma and geographical issues.

What it does

EchoCalm provides features:

  • AI-guided conversations with avatar in real-time
  • customizable and personalized avatar personality and voice to match with hallucination
  • Instant emergency button during sessions for safety affordances
  • Mood and relfection logs during post-sessions
  • Past session records tracking
  • Exportable session records in PDF for personal use or clinical healthcare
  • Help Page for navigation guidance and FAQ

How we built it

  • Frontend: React (JSX+CSS)
  • Backend: Drogon C++ Framework
  • Database: Local JSON file storage
  • Real-time Chat: WebSocket connection
  • AI simulation: OpenRouter API (meta-llama/llama-4-maverick model)
  • Extra features: TTS(text-to-speech), STT(speech-to-text) synthesis

Challenges we ran into

  • This was my first time doing full-stack website development, so I had to learn frontend and backend integration from scratch
  • Time constraints to train own AI model (so I used a hosted model for simulation instead)
  • Struggle with online deployment (so there is no hosted demo link yet, but a run-through demo video is provided)
  • Some planned features have to be postponed for future development due to limited timeframe.

Accomplishments that we're proud of

  • Built a functional AI-integrated mental health tool completely from scratch
  • Self learning all necessary skills using online materials, and developing every component independently
  • Learn how to connect backend logic with a modern frontend framework such as React
  • Implemented core functionalities in first version such as avatar chatting, voice customization and after-session reflection logs

What we learned

  • How to build and connect a C++ backend server with a React frontend
  • How to integrate AI APIs for real-time conversation
  • The importance of designing for user safety in mental health tools
  • CSS coding skills

What's next for EchoCalm

  • Add human therapist collaboration for AI-human hybrid therapy
  • Introduce CBT(cognitive behavioral therapy) exercise modules to provide additional and structured support
  • Enhance voice synthesis and avatar realism such as visual representation
  • Use a cloud-based database for multi-user access
  • Properly host the app for public accessibility

Built With

  • c++
  • css
  • drogon
  • json
  • llama4maverick
  • openrouter
  • reactjsx
  • stt
  • tts
  • websocket
Share this project:

Updates