A soulful web app that listens, remembers, heals, and grows with you.
Thera is a safe, emotionally intelligent space designed for reflection, healing, and growth. It helps users process feelings, uncover emotional patterns, engage in soothing rituals, and visually watch their personal healing journey — like tending to a garden of the soul.
Purpose: Help users feel seen and understood by identifying recurring emotional themes.
-
Journal by text or voice daily
-
AI detects emotional tones, themes (e.g., "loneliness", "shame"), and patterns
-
Patterns are labeled as Wound Seeds
-
When a wound resurfaces, Thera reflects it gently:
“You’ve mentioned loneliness on weekends a few times. Want to explore where that comes from?”
Core Stack:
TensorFlow, Whisper (voice-to-text), MongoDB, HuggingFace (RoBERTa), LangChain (retrieval)
Purpose: Provide ritual-based soul care for emotional release and grounding.
| Ritual | Action | Animation |
|---|---|---|
| Silence | Breath + inspirational quote | Fade-in quote + ripple effect |
| Release | Speak pain, watch it float away | Leaf falls, paper burns, etc. |
| Truth | Answer deep prompt | Calm typewriter animation |
| Light | Receive affirmations | Glowing word animations |
| Connection | Prompt to message a friend | Heartbeat glow pulse |
Animation Tools: Lottie, Framer Motion, GSAP
Scheduling: Node.js cron jobs + user settings
Purpose: Thera is not just an assistant. She is a calming, empathetic companion.
-
Choose your companion’s name, voice, and visual (orb, tree spirit, candle)
-
TTS responses like:
“I’m here. Let’s sit in the quiet together.”
-
Companion evolves over time (smiles, glows, blossoms)
Tech: Custom prompt engineering, TTS API, SVG + Framer Motion
Purpose: Support during emotionally intense moments.
- TensorFlow scores journal entries for emotional weight
- Distress progress bar shows increasing emotional strain
- When full:
- Quiet mode activates (e.g., breath guidance, journal)
- User can call emergency contact or therapist
“This entry feels heavy. Want to breathe together or reach out to someone?”
Alerts: Twilio, WhatsApp API
Detection: TensorFlow + custom emotion classifier
UI: React + Progress Indicators
Purpose: Show your healing as a peaceful, interactive garden.
- Each Wound Seed becomes a plant
- Rituals water it → plant sprouts → flower/fruits bloom
- Garden grows as emotional patterns are explored
- Each plant is clickable for reflective insights
Tools: Lottie, Canvas API, SVG, GSAP
Storage: MongoDB (per-user visual + wound states)
Purpose: Encourage consistent emotional check-ins.
- Streaks: Maintain rituals and journaling daily
- Rewards:
- New garden themes
- Surprise affirmations or video content
- Emotion-Based Content:
- Suggested motivational YouTube clips based on mood
| Emotion | Video Prompt |
|---|---|
| Sad | “You’re not alone” |
| Angry | “Channel it into change” |
| Hopeful | “Look how far you’ve come” |
Tools: Firebase Cloud Messaging, YouTube API, Emotion filters
Frontend: React carousel with emotional context
Here are some glimpses of Thera in action:
Chat with Thera (Soul Companion):

| Feature | Stack |
|---|---|
| Journal input (text/voice) | React, Node.js, MongoDB, Whisper |
| AI pattern detection | TensorFlow, HuggingFace (RoBERTa) |
| Daily ritual (1–2 types) | React + Lottie/GSAP |
| Emotion progress bar | React + TensorFlow classifier |
| Soul garden (basic view) | Canvas or SVG + Framer Motion |
| Emergency contact | MongoDB + optional Twilio integration |
| Notifications | Node.js or Firebase |
Frontend:
- React.js
- Framer Motion / GSAP / Lottie for animation
- Canvas or SVG for custom visualizations
Backend:
- Node.js + Express.js
- MongoDB (Mongoose for modeling)
- LangChain (retrieval + memory connection)
- Chroma (vector DB, optional)
AI & Voice:
- Hugging Face (RoBERTa) for emotion classification
- TensorFlow (custom pattern & distress detection)
- OpenAI Whisper for speech-to-text
- Teachable Machine (for lightweight emotion detection via audio)
Notifications & Alerts:
- Twilio API / WhatsApp integration
- Firebase (for mobile/web notifications)
https://github.com/Aden1ke/Thera.git
Separate Deployments: It's recommended to deploy your frontend and backend as separate services. Environment Variables: Ensure all necessary environment variables are set correctly in your deployment environment (e.g., Vercel, Netlify for frontend; Heroku, Render, AWS, DigitalOcean for backend). CORS: Adjust CORS origins in your backend (server.js and youTubeRoutes.js) to match your deployed frontend URL(s). Database: Ensure your MongoDB database is accessible from your backend deployment.
NEXT_PUBLIC_WS_ENDPOINT="ws://localhost:5000" # For local development REACT_APP_BACKEND_URL="http://localhost:5000" # For local development
DEV_DB_URI="mongodb+srv://user_dev:password_dev@dev-cluster.abcdef.mongodb.net/?retryWrites=true&w=majority&appName=dev-cluster" NODE_ENV=development OPENAI_API_KEY="sk-fakeopenaiapikey0123456789abcdefghijklmnopqrstuvwxyz" YOUTUBE_API_KEY="AIzaSyB_fakeyoutubeapikey1234567890" JWT_SECRET="FakeSuperSecretKeyForJWTsUsedInProdOnlyPleaseDontShare" FRONTEND_URL="http://localhost:3000" # For local development EMAIL_USER=example@gmail.com EMAIL_PASS=wrfg hhjj hhhuy
git clone https://github.com/your-username/thera.git
cd thera
## ⚙️ Setup
### Backend
```bash
cd backend
npm install
npm run dev
### Frontend
```bash
cd backend
npm install
npm run dev
## 💻 Usage
1. Start both frontend and backend servers.
2. Sign up / log in to your Thera account.
3. Write or speak your thoughts to Thera.
4. Thera will:
- Reflect emotional themes
- Suggest rituals
- Water your Soul Garden
- Provide soothing presence
- Send daily motivational nudges and grow over time
---
## License
This project is licensed under the MIT License.




