Transcribe β’ Summarize β’ Extract Insights β’ Chat with Meetings
Built completely with Python using FREE AI tools.
AI Meeting Assistant is a complete end-to-end meeting intelligence system that can:
- Transcribe meetings from audio, video, or YouTube links
- Support English, Hindi, and Hinglish transcription
- Generate clean AI summaries
- Extract action items, decisions, and open questions
- Enable RAG-powered chat with meetings
- Export reports as PDF or TXT
- YouTube URL support
- Upload audio/video files
- Supports:
mp3mp4wavm4amov
- Powered by OpenAI Whisper
- Runs locally
- No paid API required
- Powered by Sarvam AI
- Better regional language understanding
The assistant automatically analyzes meetings and extracts:
- β Bullet-point summaries
- β Action items
- β Owner & deadlines
- β Key decisions
- β Open questions
- β Follow-ups
Ask questions like:
What decisions were made?
Who is responsible for deployment?
What are the pending tasks?
Powered by:
- ChromaDB
- HuggingFace Embeddings
- LangChain LCEL
- Mistral AI
| Technology | Purpose |
|---|---|
| Python | Core Backend |
| Whisper AI | English Transcription |
| Sarvam AI | Hindi/Hinglish Transcription |
| LangChain LCEL | AI Pipelines |
| Mistral AI | LLM |
| ChromaDB | Vector Database |
| HuggingFace Embeddings | Local Embeddings |
| Streamlit | UI |
Input
β
Transcription
β
Meeting Analysis
β
RAG + Vector Search
β
Chat & Insights
β
Export Report
.
βββ core/
βββ utils/
βββ .gitignore
βββ project_diagram.png
βββ requirements.txt
βββ test.py
git clone <your-repo-url>
cd <project-folder>pip install -r requirements.txtstreamlit run app.pyThe assistant can export:
- π PDF Reports
- π TXT Reports
Most AI meeting assistants are expensive monthly subscriptions.
This project gives you:
- Full local processing
- Free AI stack
- RAG-powered meeting memory
- Multi-language support
- Complete ownership of your data
- Speaker Diarization
- Live Meeting Transcription
- Team Workspace
- Cloud Deployment
- Real-time Meeting Chat
- Calendar Integration
MIT License
Built with β€οΈ using Python + Open Source AI
