meeting-note

AI-powered meeting note taker for Windows & macOS.

Record system audio + microphone → Whisper transcription → AI summary → Markdown / Notion / Slack.

Zero-config system audio capture via WASAPI loopback — no Stereo Mix or virtual cables needed.

Features

• Zero-config audio capture — System audio (WASAPI loopback) + microphone via Web Audio API. No Stereo Mix or VB-Cable required.
• AI transcription — Local (faster-whisper), OpenAI Whisper API, or Remote SSH
• Multi-LLM summary — Claude Code CLI (free with subscription), Anthropic API, OpenAI GPT (Beta), Google Gemini (Beta)
• Multi-output — Save as Markdown, push to Notion, Slack, or Remote SCP
• Glassmorphism UI — Always-on-top floating bar with dark theme
• 5-step onboarding — Guided setup with dependency detection and model download
• Global hotkeys — Record, pause, stop from anywhere
• Cross-platform — Windows 10+ and macOS 12.3+

Download

Download the latest release from GitHub Releases.

Prerequisites

• FFmpeg — Required for audio format conversion
  • Windows: winget install Gyan.FFmpeg
  • macOS: brew install ffmpeg
• For local transcription:
  • Python 3.10+: pip install faster-whisper
• For summary (choose one):
  • Claude Code CLI (free with subscription): npm install -g @anthropic-ai/claude-code
  • Or any API key: Anthropic / OpenAI / Google Gemini

Quick Start

# Development
npm install
npm run dev

# Build & Package
npm run build
npm run package

The onboarding wizard guides you through all configuration on first launch.

How It Works

1. Record — Click the red button or press Ctrl+Shift+R. System audio and microphone are captured simultaneously via Web Audio API (WASAPI loopback on Windows).
2. Transcribe — Audio is converted to WAV and processed by Whisper (locally or via API).
3. Summarize — Transcript is sent to your chosen LLM for structured meeting notes.
4. Share — Notes are saved as Markdown and optionally pushed to Notion, Slack, or remote server.

Summary Providers

Provider	Cost	Setup
Claude Code CLI	Free (with $20/mo+ subscription)	Install CLI, sign in
Anthropic API	Pay-per-use	API key
OpenAI API (Beta)	Pay-per-use	API key
Google Gemini (Beta)	Pay-per-use	API key

Configuration

Settings are accessible via the gear icon (⚙) in the floating bar:

• Recording — Microphone selection (system audio is automatic)
• Transcription — Mode, language, model size
• Summary — LLM provider and API keys
• Output — Save directory
• Integrations — Notion, Slack, Remote SCP

Config file: %APPDATA%/meeting-note/config.yml (Windows) / ~/Library/Application Support/meeting-note/config.yml (macOS)

Hotkeys

Windows	macOS	Action
Ctrl+Shift+R	Cmd+Shift+R	Start recording
Ctrl+Shift+P	Cmd+Shift+P	Pause / Resume
Ctrl+Shift+S	Cmd+Shift+S	Stop & process
Ctrl+Shift+M	Cmd+Shift+M	Show / Hide

Architecture

src/
├── main/              # Electron main process
│   ├── index.ts       # Window, tray, hotkeys, WASAPI loopback setup
│   ├── recorder.ts    # Audio buffer save + WebM→WAV conversion
│   ├── transcriber.ts # Whisper (local/API/SSH)
│   ├── summarizer.ts  # Multi-LLM (Claude CLI/Anthropic/OpenAI/Gemini)
│   ├── pipeline.ts    # Orchestration with progress tracking
│   └── publishers/    # Markdown, Notion, Slack, SCP
├── preload/           # Context bridge (IPC API)
├── renderer/          # React + Tailwind UI
│   ├── components/    # FloatingBar, Onboarding, Settings, etc.
│   └── hooks/         # useRecording, useAudioRecorder, useConfig
└── shared/            # Zod config schema

Contributing

Pull requests welcome. The project uses GitHub Actions CI for build, lint, and type checking on every PR.

npm run build    # Build all targets
npm run lint     # ESLint
npm run package  # Create installer

License

MIT