🎙️ MeetingMind

A 100% free, offline meeting notes assistant that captures system audio from Teams/Zoom calls, transcribes with speaker identification, and generates intelligent summaries with action items - all running locally on your machine!

---

📥 Quick Download

Platform	Download	Requirements
Windows	📦 Download EXE	Windows 10/11, Ollama, FFmpeg

First Time Setup

1. Download the latest release ZIP
2. Extract to any folder
3. Install Ollama from https://ollama.ai
4. Open terminal and run: ollama pull llama3.2
5. Run MeetingMind.exe
6. Done! 🎉

💡 See INSTALL.md for detailed setup instructions

---

✨ Features

🎤 Audio Capture

• System Audio Recording - Capture audio from Teams, Zoom, or any application (Windows WASAPI loopback)
• Auto-Detect Meetings - Automatically start recording when Teams/Zoom/Meet opens
• Real-Time Transcription - See text as you speak during meetings ⚡ NEW
• Upload Support - Drag & drop existing audio/video files
• System Tray - Quick recording controls from your taskbar

🎯 Transcription & Speaker ID

• Accurate Transcription - Uses OpenAI Whisper (runs 100% offline)
• Speaker Diarization - Identifies who's speaking using pyannote-audio
• Audio Snippets - Listen to clips to identify unknown speakers
• Word-Level Timestamps - Precise timing for each word ⚡ NEW

🌟 Highlights & Bookmarks ⚡ NEW

• Manual Bookmarks - Mark important moments with Ctrl+Shift+B
• Auto-Detection - AI identifies action items, decisions, questions
• Clip Extraction - Export audio clips for specific highlights
• Quick Navigation - Jump to highlighted moments

🤖 Smart Q&A Workflow

• Speaker Identification - Audio clips help you name each speaker
• Clarifying Questions - AI asks questions to improve note accuracy
• Quick or Detailed Mode - Choose 3-5 or 5-10 questions

📋 Intelligent Summaries

• Smart Summaries - Powered by Ollama LLMs (llama3.2, mistral, etc.)
• Meeting Templates - Standup, planning, retrospective, 1:1 templates
• Action Items - Automatically extracts tasks with assignees
• Key Decisions - Highlights important decisions made

💬 Meeting Chat ⚡ NEW

• Ask Questions - Query your past meetings with AI
• Semantic Search - Find meetings by meaning, not just keywords
• Context-Aware - Understands follow-up questions
• Cross-Meeting Insights - Find patterns across meetings

📊 Analytics Dashboard ⚡ NEW

• Meeting Statistics - Duration, participants, word count
• Speaker Analytics - Talk time per person
• Meeting Trends - Patterns by day, week, month
• Productivity Score - Based on action items and decisions

🔗 Integrations ⚡ NEW

• Slack - Send notes via webhook
• Microsoft Teams - Share to Teams channels
• Notion - Export to Notion databases
• Email - Send via SMTP
• Calendar Sync - Google Calendar & Outlook

⌨️ Keyboard Shortcuts ⚡ NEW

• Ctrl+Shift+R - Toggle recording
• Ctrl+Shift+B - Add bookmark
• Ctrl+Shift+A - Mark action item
• Ctrl+Shift+D - Mark decision

📥 Export Options

• Markdown - Clean, portable format
• HTML - Styled web pages
• JSON - For integrations
• DOCX - Word documents
• PDF - Professional reports

📚 Meeting History

• Search & Browse - Find past meetings by date, participant, or keyword
• Statistics - Track meeting duration and frequency
• Quick Reload - Re-open and export any saved meeting

🔒 Privacy First

• 100% Offline - Everything runs locally, no API calls
• No Data Collection - Your meetings stay on your machine
• Completely Free - No API costs, no subscriptions

---

📋 Requirements

• Windows 10/11 (for system audio capture)
• Ollama (local LLM runtime) - Download
• FFmpeg (for audio processing) - Download
• 8GB+ RAM recommended (16GB for larger models)

---

🚀 Installation

Option A: Download Pre-built EXE (Easiest!)

1. Go to Releases
2. Download MeetingMind-Windows.zip
3. Extract the ZIP file
4. Install Ollama and run ollama pull llama3.2
5. Double-click MeetingMind.exe
6. Your browser opens to http://localhost:7860

Option B: Run from Source (For Developers)

1. Install FFmpeg

Windows:

# Using Chocolatey
choco install ffmpeg

# Or download from: https://ffmpeg.org/download.html

macOS:

brew install ffmpeg

Linux:

sudo apt install ffmpeg  # Ubuntu/Debian
sudo dnf install ffmpeg  # Fedora

2. Install Ollama

Download and install from: https://ollama.ai

After installation, pull a model:

ollama pull llama3.2
# or
ollama pull mistral

Start Ollama server:

ollama serve

💡 Keep the Ollama server running while using MeetingMind!

3. Clone/Download MeetingMind

cd MeetingMind

4. Install Python Dependencies

# Create a virtual environment (recommended)
python -m venv venv

# Activate it
.\venv\Scripts\Activate.ps1  # Windows PowerShell
# or
source venv/bin/activate      # macOS/Linux

# Install packages
pip install -r requirements.txt

⏱️ First install may take 5-10 minutes (Whisper models are ~150MB)

5. Run the App

python main.py

The app will open automatically in your browser at http://127.0.0.1:7860

A system tray icon will appear for quick recording controls!

🎯 How to Use

Recording a Meeting

1. Start Recording

   • Click "🔴 Start Recording" in the web UI, OR
   • Right-click the system tray icon → "Start Recording"

2. During the Meeting

   • Join your Teams/Zoom call as normal
   • MeetingMind captures all system audio

3. Stop Recording

   • Click "⏹️ Stop Recording" when done
   • Processing starts automatically

Q&A Workflow (Key Feature!)

After processing, you'll be guided through a Q&A session:

1. Speaker Identification

   • Listen to audio clips of each speaker
   • Enter their name (e.g., "John Smith")
   • This improves meeting notes accuracy

2. Clarifying Questions

   • AI asks about unclear action items, decisions
   • Answer or skip questions as needed
   • Choose "Skip All" to generate notes immediately

3. Review Results

   • Executive summary
   • Key points and decisions
   • Action items with assignees

Uploading Existing Audio

1. Go to "📝 New Meeting" tab
2. Upload your audio/video file (mp3, wav, m4a, mp4, etc.)
3. Click "Process Upload"
4. Complete Q&A workflow
5. Export results

📁 Project Structure

MeetingMind/
├── main.py                 # Main entry point
├── app.py                  # Alternative entry (runs Gradio only)
├── build.bat               # Windows EXE build script
├── build.py                # Python build script
├── meetingmind.spec        # PyInstaller configuration
├── core/
│   ├── config.py          # Configuration management
│   ├── controller.py      # Main app controller (orchestrates all services)
│   └── events.py          # Event system for async communication
├── services/
│   ├── audio_capture.py   # WASAPI system audio recording
│   ├── transcriber.py     # Whisper transcription
│   ├── diarizer.py        # Speaker diarization
│   ├── summarizer.py      # Ollama summarization
│   ├── qa_engine.py       # Q&A generation & management
│   ├── meeting_detector.py # Auto-detect Teams/Zoom/Meet
│   ├── templates.py       # Meeting templates (standup, planning, etc.)
│   ├── exporter.py        # Export to MD/HTML/JSON/DOCX/PDF
│   └── history.py         # Meeting history storage & search
├── ui/
│   ├── gradio_app.py      # Web interface
│   └── system_tray.py     # System tray application
├── data/
│   ├── meetings/          # Saved meeting notes
│   └── profiles/          # Speaker profiles
├── assets/                 # Icons and images
├── docs/                   # Documentation
├── requirements.txt
└── README.md

⚙️ Configuration

Command Line Options

python main.py --help

Options:
  --port, -p    Port to run web UI (default: 7860)
  --share       Create public share link
  --no-tray     Disable system tray icon
  --no-browser  Don't open browser automatically
  --check       Check dependencies and exit

Q&A Modes

In Settings tab:

• Quick Mode: 3-5 questions (faster)
• Detailed Mode: 5-10 questions (more accurate)

Whisper Models (Speed vs Accuracy)

Model	Size	Speed	Accuracy	Recommended For
tiny	39 MB	⚡⚡⚡⚡	⭐⭐	Quick tests
base	74 MB	⚡⚡⚡	⭐⭐⭐	Most users
small	244 MB	⚡⚡	⭐⭐⭐⭐	Better accuracy
medium	769 MB	⚡	⭐⭐⭐⭐⭐	Professional use
large	1550 MB	🐌	⭐⭐⭐⭐⭐	Maximum accuracy

Ollama Models

Popular choices:

• llama3.2 (3B) - Fast, good quality ⚡ (Recommended)
• llama3.1 (8B) - Better quality, slower
• mistral (7B) - Good balance

Install with: ollama pull <model-name>

🔧 Troubleshooting

"Ollama not accessible" error

• Make sure Ollama is running: ollama serve
• Check if model is installed: ollama list
• Pull model if needed: ollama pull llama3.2

"FFmpeg not found" error

• Install FFmpeg (see installation steps above)
• Restart terminal after installation

No audio being captured

• Make sure you're on Windows 10/11
• Check audio device in Settings
• Try refreshing audio devices

Slow transcription

• Use smaller Whisper model (tiny or base)
• For GPU acceleration, install CUDA-enabled PyTorch:

  pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121

Memory issues

• Use smaller models (tiny Whisper + llama3.2)
• Close other applications
• Process shorter audio files

🎨 Supported Audio Formats

• MP3, WAV, M4A, MP4 (audio extracted)
• WEBM, OGG, FLAC
• Any format supported by FFmpeg

💡 Tips

• First-time users: Start with base Whisper model and llama3.2
• Speaker ID: Clearer audio = better speaker separation
• Long meetings: Works best with meetings under 2 hours
• Multiple languages: Whisper auto-detects language
• Background noise: Better audio = better results

🤝 Contributing

See CONTRIBUTING.md for guidelines.

📜 License

MIT License - Free to use, modify, and distribute!

🙏 Credits

Built with:

• OpenAI Whisper - Speech recognition
• Ollama - Local LLM runtime
• Gradio - UI framework

📧 Support

Having issues? Check:

1. This README
2. GitHub Issues
3. Ollama documentation: https://ollama.ai/docs

---

Made with ❤️ for productive meetings | 100% Free | 100% Offline | 100% Private