notion-voice-notes

CLI tool to transcribe audio files and create Notion pages with AI summaries

Transform voice memos into organized Notion pages with automatic transcription, AI-powered summarization, key points extraction, and action item identification.

Features

• Fast Transcription via Groq (free, fast) or OpenAI Whisper
• AI Summarization with Claude (titles, summaries, key points, action items)
• Automatic Notion Pages with structured content blocks
• Large File Support with automatic chunking for files over 25MB
• JSON Output for automation pipelines (n8n, Pipedream, etc.)
• Standalone Binary - no runtime dependencies needed

Installation

Download Binary (Recommended)

Download the pre-built binary for your platform from GitHub Releases:

# macOS
curl -L https://github.com/keithce/notion-voice-notes/releases/latest/download/voice-to-notion -o voice-to-notion
chmod +x voice-to-notion

# Linux
curl -L https://github.com/keithce/notion-voice-notes/releases/latest/download/voice-to-notion-linux -o voice-to-notion
chmod +x voice-to-notion

# Windows
# Download voice-to-notion-windows.exe from the releases page

Using Bun (Development)

git clone https://github.com/keithce/notion-voice-notes.git
cd notion-voice-notes
bun install
bun run start -- ./recording.mp3

Configuration

Create a .env file or set these environment variables:

Variable	Required	Description
ANTHROPIC_API_KEY	Yes	Anthropic API key for Claude summarization
NOTION_API_KEY	Yes	Notion integration token
NOTION_DATABASE_ID	Yes	Target Notion database ID
GROQ_API_KEY	If using Groq	Groq API key (default provider)
OPENAI_API_KEY	If using OpenAI	OpenAI API key for Whisper
TRANSCRIPTION_PROVIDER	No	Default provider: groq or openai

Notion Setup

1. Create a Notion integration
2. Share your target database with the integration
3. Copy the database ID from the database URL

Usage

# Basic usage (uses Groq by default)
voice-to-notion ./recording.mp3

# Use OpenAI Whisper instead
voice-to-notion ./recording.mp3 -t openai

# Custom title
voice-to-notion ./recording.mp3 --title "Meeting Notes - Feb 2"

# Dry run (process without creating Notion page)
voice-to-notion ./recording.mp3 --dry-run -v

# JSON output for automation
voice-to-notion ./recording.mp3 --json

Options

-t, --transcription     Transcription provider: "groq" | "openai" (default: groq)
--title                 Custom title for Notion page (default: AI-generated)
--database-id           Override Notion database ID
--dry-run               Process but don't create Notion page
-v, --verbose           Enable verbose logging to stderr
--json                  Output result as JSON to stdout
-h, --help              Show help message
--version               Show version number

n8n / Automation Integration

Use --json flag to get structured output for parsing in automation tools:

{
  "success": true,
  "audioFile": "./recording.mp3",
  "transcription": {
    "text": "Full transcription...",
    "duration": 125.5,
    "provider": "groq"
  },
  "summary": {
    "title": "AI-Generated Title",
    "summary": "Brief summary...",
    "mainPoints": ["Point 1", "Point 2"],
    "actionItems": ["Action 1", "Action 2"],
    "transcript": "Full text..."
  },
  "notion": {
    "pageId": "abc123",
    "url": "https://notion.so/..."
  }
}

Docker Usage

FROM oven/bun:latest

# Download the binary
RUN curl -L https://github.com/keithce/notion-voice-notes/releases/latest/download/voice-to-notion-linux -o /usr/local/bin/voice-to-notion \
    && chmod +x /usr/local/bin/voice-to-notion

# Your automation setup...

Building from Source

# Build for current platform
bun run build

# Build for all platforms
bun run build:all

# Outputs to dist/
#   voice-to-notion         (macOS)
#   voice-to-notion-linux   (Linux)
#   voice-to-notion-windows.exe (Windows)

Credits

This project was inspired by Thomas J Frank's excellent work:

• Original Implementation: pipedream-notion-voice-notes
• Tutorial: How to Transcribe Audio to Text with ChatGPT and Notion

Why this CLI exists: The Pipedream implementation stopped working, and recreating the workflow entirely in n8n proved difficult due to the complexity of handling audio chunking, multiple API calls, and Notion formatting. This CLI provides a standalone binary that handles all the processing internally and can be called from any automation tool.

License

MIT