MeetScribe 🎙️

AI-powered meeting assistant for Google Meet — transcribe, summarize, and analyze your meetings automatically.

MeetScribe is a full-stack application consisting of a Chrome extension that captures Google Meet transcripts in real time, a Node.js/Express backend that processes them with Gemini AI, and a Next.js frontend where users can review summaries, track action items, and explore an analytics dashboard.

---

Features

• 🎤 Live transcription — captures speech from Google Meet via Chrome extension
• 🤖 AI summarization — Gemini AI generates clean summaries, strips filler words, and deduplicates repeated phrases
• ✅ Action item extraction — automatically identifies tasks and owners from the transcript
• 👥 Participant detection — extracts participant names mentioned in the conversation
• 😊 Sentiment analysis — classifies meeting tone as positive, neutral, or negative with a score and insight
• 🏷️ Auto-tagging — generates smart tags like sprint-planning, design, urgent from the summary
• 📅 Next meeting detection — extracts any mentioned follow-up date or time
• 📊 Analytics dashboard — tracks meeting duration, participant activity, sentiment trends, and tag-based filtering
• 📤 Markdown export — download any meeting as a formatted .md file
• 🔐 Auth — email/password signup + Google OAuth via NextAuth

---

Tech Stack

Layer	Tech
Chrome Extension	Vanilla JS (content.js, background.js, popup.js)
Frontend	Next.js 14 (App Router), Tailwind CSS
Backend	Node.js, Express
Database	MongoDB (Mongoose)
AI	Google Gemini API
Auth	JWT + NextAuth (Google OAuth)

---

Project Structure

meetscribe/
├── extension/              # Chrome extension
│   ├── content.js          # Injects into Google Meet, captures transcript
│   ├── background.js       # Service worker, auth bridge
│   ├── popup.js            # Extension popup UI
│   ├── authBridge.js       # Syncs auth token between extension and web app
│   ├── authSync.js
│   └── manifest.json
│
├── backend/                # Express API
│   ├── models/
│   │   ├── SummaryModel.js # Meeting schema (transcript, summary, sentiment, tags, participants...)
│   │   └── UserModel.js
│   ├── routers/
│   │   ├── aiRouter.js     # /summarize, /action-items, /sentiment, /auto-tag, /ask, /export
│   │   ├── SumRouter.js
│   │   └── UserRouter.js   # /register, /authenticate, /google-login, /me, /history
│   ├── gemini.js           # Gemini AI wrapper
│   ├── connection.js       # MongoDB connection
│   └── index.js
│
└── frontend/               # Next.js app
    └── src/app/
        ├── components/
        │   ├── NavBar.jsx       # Responsive nav, shows Dashboard + History only when logged in
        │   └── SessionWrapper.jsx
        ├── dashboard/
        │   └── page.jsx         # Analytics dashboard (auth-protected, client-side guard)
        ├── history/
        │   └── page.jsx
        ├── login/
        ├── signup/
        └── contactus/

---

Getting Started

Prerequisites

• Node.js 18+
• MongoDB (local or Atlas)
• Google Gemini API key
• Google OAuth credentials (for NextAuth)

1. Clone the repo

git clone https://github.com/lavanya28007/meetscribe.git
cd meetscribe

2. Backend

cd backend
npm install

Create a .env file:

MONGO_URI=your_mongodb_connection_string
JWT_SECRET=your_jwt_secret
GEMINI_API_KEY=your_gemini_api_key
PORT=5000

node index.js

3. Frontend

cd frontend
npm install

Create a .env.local file:

NEXT_PUBLIC_API_URL=http://localhost:5000
NEXTAUTH_URL=http://localhost:3000
NEXTAUTH_SECRET=your_nextauth_secret
GOOGLE_CLIENT_ID=your_google_client_id
GOOGLE_CLIENT_SECRET=your_google_client_secret

npm run dev

4. Chrome Extension

1. Open Chrome and go to chrome://extensions
2. Enable Developer mode
3. Click Load unpacked and select the extension/ folder
4. Open Google Meet — MeetScribe will activate automatically

---

API Reference

All AI routes are under /ai, user routes under /user.

AI Routes

Method	Endpoint	Description
POST	/ai/summarize	Summarize a transcript, extract participants + next meeting
POST	/ai/action-items	Extract action items from a transcript
POST	/ai/sentiment	Analyze meeting sentiment (label, score, insight)
POST	/ai/auto-tag	Generate smart tags from a summary
POST	/ai/ask	Ask a question about a specific meeting
PATCH	/ai/tag/:summaryId	Manually add/remove tags, star, or archive a meeting
GET	/ai/export/:summaryId	Download meeting as Markdown

User Routes

Method	Endpoint	Auth	Description
POST	/user/register	—	Register with email/password
POST	/user/authenticate	—	Login, returns JWT
POST	/user/google-login	—	Google OAuth login
GET	/user/me	✅	Get current user profile
GET	/user/history	✅	Get all meetings for the logged-in user
DELETE	/user/history/:id	✅	Delete a meeting

---

Data Model

Each meeting (Summary) stores:

{
  meetingTitle: String,
  transcript: String,
  summary: String,
  userId: ObjectId,          // ref: users
  actionItems: [String],
  participants: [String],    // auto-extracted from transcript
  sentiment: {
    label: "positive" | "neutral" | "negative",
    score: Number,           // -1.0 to 1.0
    insight: String,
    breakdown: [{ section, tone }]
  },
  tags: [String],            // auto-generated or manually added
  starred: Boolean,
  archived: Boolean,
  duration: Number,          // minutes
  nextMeeting: String,
  exportedAt: Date,
  createdAt: Date
}

---

Analytics Dashboard

The dashboard (/dashboard) is protected — only accessible to logged-in users. It shows:

• Total meetings, average duration, average participants, positive sentiment %
• Sentiment breakdown bar (positive / neutral / negative)
• Top participants ranked by meeting frequency
• Filterable meetings table (by time range, sentiment, title search)
• Click-through drawer with full meeting details

Auth is handled client-side: the page reads the JWT from localStorage on mount and redirects to /login if no valid session is found.

---

Contributing

Pull requests are welcome. For major changes, open an issue first to discuss what you'd like to change.

---

License

MIT