MeetingMind AI

Automatically transcribe meeting recordings and generate structured minutes using AI. Upload an audio file, and the app returns a title, summary, key points, action items, identified participants, and the full transcript — all in seconds.

Built with FastAPI + PostgreSQL on the backend and React + Material-UI on the frontend. Runs fully containerised with Docker Compose.

---

Features

• Drag-and-drop audio upload — MP3, WAV, M4A, OGG, FLAC supported
• File size limit — up to 25 MB; FFmpeg is available on the backend to compress if needed
• AI transcription — OpenAI Whisper API (whisper-1)
• Structured extraction — GPT-4o-mini returns title, summary, key points, action items, and participants as structured JSON
• Tabbed meeting detail view — Summary / Key Points / Action Items / Transcript
• Real-time processing status — progress bar polling while the background task runs
• Trello integration — push meeting minutes to a Trello card with one click
• Pagination — meeting list paginates at 5 per page
• Delete meetings — removes the database record and the stored audio file

---

Architecture

Browser (React + MUI)
       │  REST / JSON
       ▼
FastAPI (Python 3.11)
  ├── POST /meeting/upload-audio        → saves file, starts background task
  ├── GET  /meeting/processing-status   → polls task progress
  ├── GET  /meeting/get_meetings        → paginated list
  ├── GET  /meeting/get_meeting_by_id   → full detail + transcript
  ├── DELETE /meeting/delete_meeting    → removes record + audio file
  └── GET  /meeting/send_to_trello      → creates Trello card
       │
       ├── OpenAI Whisper API  (transcription)
       ├── OpenAI GPT-4o-mini  (structured extraction)
       └── PostgreSQL          (meetings + transcriptions)

Background task flow:

Upload → Save audio → [FFmpeg compress if >24 MB] → Whisper API
       → Store transcript → GPT-4o-mini extraction → Store meeting → Done

---

Project Structure

ai-meetings-minutes/
├── docker-compose.yml          # Production: postgres + backend + nginx frontend
├── docker-compose.dev.yml      # Development: hot-reload backend + React dev server
│
├── backend/
│   ├── Dockerfile
│   ├── requirements.txt
│   ├── .env.example            # Copy to .env and fill in your keys
│   ├── main.py                 # FastAPI app, CORS, router registration
│   ├── config.py               # Settings loaded from .env
│   ├── database.py             # SQLAlchemy engine + session factory
│   ├── models/
│   │   └── models.py           # Meeting, Transcription, Trello ORM models
│   ├── schemas/
│   │   ├── meetingSchema.py
│   │   └── transcriptionSchema.py
│   ├── routers/
│   │   └── meeting.py          # Route definitions
│   ├── controller/
│   │   └── meetingController.py  # Business logic, AI calls, file handling
│   └── recordings/             # Audio files stored as recordings/YYYY/MM/<uuid>/
│
└── frontend/
    ├── Dockerfile              # Multi-stage: node build → nginx serve
    ├── nginx.conf
    ├── src/
    │   ├── App.tsx             # Theme, routing, NavBar
    │   ├── services/
    │   │   └── api.ts          # All API calls in one place
    │   └── components/
    │       ├── Home.tsx        # Drag-and-drop upload page
    │       ├── MeetingList.tsx # Paginated list with live processing status
    │       └── MeetingDetails.tsx  # Tabbed detail view + Trello button
    └── public/

---

Quick Start (Docker)

1. Clone and configure

git clone <repo-url>
cd ai-meetings-minutes
cp backend/.env.example backend/.env

Edit backend/.env:

POSTGRES_USER=postgres
POSTGRES_PASSWORD=yourpassword
POSTGRES_SERVER=localhost
POSTGRES_PORT=5432
POSTGRES_DB=ai_meeting_minutes

OPENAI_API_KEY=sk-...

# Optional — only needed for Trello integration
TRELLO_API_KEY=
TRELLO_API_TOKEN=
TRELLO_LIST_ID=

2. Start production

docker compose up --build

Service	URL
Frontend	http://localhost:3000
Backend	http://localhost:8000
API docs	http://localhost:8000/docs

3. Start dev mode (hot reload on save)

docker compose -f docker-compose.dev.yml up --build

• Editing any .py file → uvicorn restarts automatically
• Editing any .tsx/.ts file → React dev server hot-reloads in the browser

---

Manual Setup (without Docker)

Prerequisites

• Python 3.11+
• Node.js 20+
• PostgreSQL 14+
• FFmpeg (brew install ffmpeg / apt install ffmpeg / ffmpeg.org)

Backend

cd backend
python -m venv venv

# Windows
.\venv\Scripts\activate
# Linux / Mac
source venv/bin/activate

pip install -r requirements.txt
cp .env.example .env   # then fill in your values
uvicorn main:app --reload
# → http://localhost:8000

Frontend

cd frontend
npm install
# create .env with:  REACT_APP_API_URL=http://localhost:8000/meeting
npm start
# → http://localhost:3000

---

Environment Variables

Variable	Required	Description
POSTGRES_USER	✅	PostgreSQL username
POSTGRES_PASSWORD	✅	PostgreSQL password
POSTGRES_SERVER	✅	Host (localhost or db in Docker)
POSTGRES_PORT	✅	Default 5432
POSTGRES_DB	✅	Database name
OPENAI_API_KEY	✅	Used for Whisper transcription + GPT-4o-mini
TRELLO_API_KEY	❌	Trello Power-Up key
TRELLO_API_TOKEN	❌	Trello OAuth token
TRELLO_LIST_ID	❌	Target list ID for new cards

---

API Reference

Method	Endpoint	Description
POST	/meeting/upload-audio	Upload audio; returns task_id
GET	/meeting/processing-status/{task_id}	Poll processing progress
GET	/meeting/get_meetings	List meetings (skip, limit)
GET	/meeting/get_meeting_by_id/{id}	Full meeting detail + transcript
DELETE	/meeting/delete_meeting/{id}	Delete meeting + audio file
GET	/meeting/send_to_trello/{id}	Push minutes to Trello

Interactive docs available at /docs (Swagger) and /redoc when the backend is running.

---

Tech Stack

Layer	Technology
Frontend	React 19, TypeScript, Material-UI v7
Backend	FastAPI, Python 3.11, Uvicorn
Database	PostgreSQL 16, SQLAlchemy 2
AI	OpenAI Whisper API, GPT-4o-mini
Audio	FFmpeg (auto-compression)
Containers	Docker, Docker Compose
Prod server	nginx (static + API proxy)

---

License

MIT