Speak. Listen. Conquer.
An AI-powered voice assistant for the web. Paste any URL, click the mic, speak a command — and the AI responds out loud.
Built as a full-stack project by a first-year engineering student. No frameworks. No shortcuts. Just raw skill.
🔗 Coming soon after deployment in Phase 7
Add screenshots here after completing Phase 6 polish
| Login Page | Main App | Dashboard |
|---|---|---|
screenshot-login.png |
screenshot-app.png |
screenshot-dashboard.png |
| User | How It Helps |
|---|---|
| 👁️ Blind / Visually Impaired | Hear webpage content without reading |
| 👴 Elderly Users | Simplify complex websites with voice |
| 🖐️ Motor-Impaired Users | Navigate completely hands-free |
| 🎓 Busy Students | Summarize long pages in seconds |
| 🌍 Non-native Speakers | Translate & simplify page content |
| Layer | Technology | Purpose |
|---|---|---|
| Frontend | HTML5 + CSS3 | Structure, dark/light UI, animations |
| Logic | Vanilla JavaScript | Events, DOM, API calls |
| Voice In | Web Speech API | Browser microphone — free, no setup |
| Voice Out | SpeechSynthesis API | AI talks back out loud |
| Animations | GSAP | Smooth, pro-level motion |
| Backend | Node.js + Express.js | Server + REST API endpoints |
| AI Brain | Groq API (LLaMA 3.3 70B) | Summarize, simplify, translate |
| Database | Firebase Firestore | Save query history per user |
| Auth | Firebase Auth | Email + Google sign-in |
| Frontend Deploy | Vercel | Live URL, free tier |
| Backend Deploy | Railway | Node.js hosting, free tier |
| Version Control | Git + GitHub | Code history, portfolio proof |
TALK-AI-TIVE/
├── FrontEnd/
│ ├── index.html ← Login / Signup page
│ ├── app.html ← Main AI voice assistant
│ ├── dashboard.html ← Query history & analytics
│ ├── style.css ← Shared styles
│ └── app.js ← Voice + AI logic
├── BackEnd/
│ ├── server.js ← Express server entry point
│ ├── firebase.js ← Firebase Admin SDK setup
│ ├── claude.js ← Groq AI API integration
│ ├── routes/
│ │ └── claude.js ← /api/ask route handler
│ └── .env ← Secret keys (never committed)
├── .gitignore
├── BLUEPRINT
└── README.md
| Phase | Name | Status |
|---|---|---|
| 🟥 Phase 1 | Login Page HTML Structure | ✅ Done |
| 🟧 Phase 2 | Main App UI + Full CSS Dark Theme | ✅ Done |
| 🟨 Phase 3 | Voice Input + Dashboard Page | ✅ Done |
| 🟩 Phase 4 | Backend + Groq AI API Integration | ✅ Done |
| 🟩 Phase 4.5 | Frontend ↔ Backend Connected + Live AI Responses | ✅ Done |
| 🔵 Phase 5 | Firebase Auth + Firestore History Saving | ✅ Done |
| 🏁 Phase 6 | GSAP Animations + Dashboard + UX Polish | ⏳ Up Next |
| 🚀 Phase 7 | Deployment (Vercel + Railway) | ⏳ Upcoming |
- Express.js server running on port 5000
- Groq API (LLaMA 3.3 70B) successfully integrated
/api/askendpoint live and tested — takes a URL + voice command → scrapes the page → returns AI response- Web scraping with axios + cheerio working
- Tested with real Wikipedia URLs — sub-2 second AI responses confirmed ⚡
- Firebase Auth — Email + Google sign-in working ✅
- Firestore Database — Every voice query + AI response saved to user history ✅
/api/historyPOST endpoint — saves query to Firestore ✅/api/history/:userIdGET endpoint — retrieves user's past queries ✅.envsecured,node_modulesand secrets gitignored ✅
- Node.js (LTS)
- Git
- VS Code + Live Server extension
git clone https://github.com/Max-Planxk/TALK-AI-TIVE.git
cd TALK-AI-TIVEOpen FrontEnd/index.html with Live Server in VS Code.
cd BackEnd
npm installCreate a .env file inside BackEnd/:
GROQ_API_KEY=your_groq_api_key_here
PORT=5000
Start the server:
node server.js# In a second terminal (PowerShell)
Invoke-WebRequest -Uri "http://localhost:5000/api/ask" -Method POST -Headers @{"Content-Type"="application/json"} -Body '{"url":"https://en.wikipedia.org/wiki/Artificial_intelligence","userCommand":"what is this page about","mode":"summarize"}'| Mode | Voice Command Example | What AI Does |
|---|---|---|
| 🔍 Summarize | "Summarize this page" | Gives a concise summary of the URL content |
| 🧒 Simplify | "Explain this like I'm 5" | Breaks it down in plain simple language |
| 🌍 Translate | "Translate this to Hindi" | Translates page content to Hindi |
| Method | Endpoint | Status | Description |
|---|---|---|---|
POST |
/api/ask |
✅ Live | Send URL + command → get AI response |
POST |
/api/history |
✅ Live | Save a query + response to Firestore |
GET |
/api/history/:userId |
✅ Live | Get user's past queries from Firestore |
DELETE |
/api/history/:id |
⏳ Phase 6 | Delete a specific query from history |
Create a .env file in BackEnd/ — never commit this file.
GROQ_API_KEY= # From console.groq.com — free
PORT=5000
Firebase credentials are handled via the Firebase Admin SDK service account — also never committed.
- Built TALK-AI-TIVE, a full-stack AI voice assistant web app using Groq API (LLaMA 3.3 70B), Web Speech API, Node.js, and Firebase — enabling users to interact with any webpage hands-free.
- Implemented real-time voice recognition and AI-powered webpage summarization using Groq's LLaMA API, achieving sub-2 second response times.
- Designed and deployed a full-stack accessibility tool with Firebase authentication, Firestore database, Express.js backend, and Vercel/Railway hosting.
HTML5 · CSS3 · Flexbox & Grid · Glassmorphism UI · JavaScript DOM · Fetch API · Web Speech API · Node.js · Express.js · REST APIs · Groq API · Web Scraping · Firebase Auth · Firestore · Git & GitHub · Vercel · Railway · GSAP Animations
🧠 Understand every line before moving to the next.
⌨️ Type the code yourself — never copy paste blindly.
🔧 Break things. Fix things. That is how you learn.
🔍 Google every error message — this is a real developer skill.
📦 Commit to GitHub after every phase completes.
🤝 If stuck for more than 30 minutes — ask for help.
👑 The goal is not just to finish. The goal is to OWN it.
MIT — free to use, learn from, and build on.
"My fear of being mediocre is greater than my fear of dying."
Built with 🔥 by a first-year engineer who refused to be average.