Summarize MCP

🤖Co-authored withClaude Code- Making AI summaries audible since 2025! 🔊

git clone https://github.com/FiveOhhWon/summarize-mcp.git cd summarize-mcp pip install -e .

{ "mcpServers": { "summarize": { "command": "python", "args": ["/absolute/path/to/summarize-mcp/src/summarize_mcp/server.py"], "env": { "OPENAI_API_KEY": "your-openai-api-key" } } } }

- OPENAI_API_KEY(required): Your OpenAI API key
- DEBUG(optional): Set to"true"for verbose logging

Converts text to speech and plays it in the background. Uses saved voice and tone preferences unless overridden.

- summary(required): The text to convert to speech
- voice(optional): Voice to use -alloy,ash,ballad,coral,echo,fable,nova,onyx,sage, orshimmer(uses saved preference if not specified)
- instructions(optional): Instructions for how the text should be spoken (uses saved tone if not specified)

{ "summary": "The quick brown fox jumps over the lazy dog. This pangram contains all letters of the alphabet.", "voice": "nova", "instructions": "Speak slowly and clearly, emphasizing each word." }

Set the default voice for all future text-to-speech conversions.

- voice(required): The voice to use -alloy,ash,ballad,coral,echo,fable,nova,onyx,sage, orshimmer

Set the default tone/instructions for how text should be spoken in all future TTS requests.

- tone(required): The tone/instructions to use (e.g., "Speak slowly and calmly", "Be enthusiastic and energetic")

{ "tone": "Speak in a warm, friendly manner with moderate pacing" }

"Please summarize this article and play it as audio"

- Generate a summary of the content
- Use theplay_summarytool to convert it to speech
- The audio will play in the background with saved preferences

"Set the default voice to nova"

This will save "nova" as your preferred voice for all future summaries.

"Set the tone to be warm and conversational with a slower pace"

This will save your tone preference for all future summaries.

"Summarize this document and play it using the 'sage' voice"

This will use "sage" for this summary only, without changing your default.

"Create an audio summary of this text. Make it sound enthusiastic and energetic."

This will use custom instructions for this summary only.

# Install dependencies pip install -r requirements.txt # Install in development mode pip install -e . # Run the server python -m summarize_mcp # Run tests python test.py # Run with debug logging DEBUG=true python -m summarize_mcp

summarize-mcp/ ├── src/ │ └── summarize_mcp/ │ ├── __init__.py # Package initialization │ ├── __main__.py # Entry point for python -m │ └── server.py # Main MCP server implementation ├── pyproject.toml # Python project metadata ├── requirements.txt # Python dependencies ├── test.py # Test script └── README.md # This file

- Audio Format: MP3 (OpenAI TTS output format)
- Temporary Files: Stored in system temp directory
- File Cleanup: Automatic cleanup after 10 seconds (configurable)
- Old File Purge: Files older than 1 hour are cleaned on startup
- Platform Support:

- macOS: Uses built-inafplay
- Windows: Uses PowerShell with Windows Media Player
- Linux: Auto-detects available player (mpg123, sox, ffmpeg, vlc, alsa)
- Fallback: Opens with system default audio application

- Preferences saved to~/.summarize-mcp-state.json
- Persists voice and tone settings between sessions
- Automatic loading on startup

"OPENAI_API_KEY environment variable is not set"

Set your OpenAI API key in the Claude Desktop configuration.

Linux users: Install one of the supported audio players:

# Ubuntu/Debian sudo apt-get install mpg123 # or sudo apt-get install sox # or sudo apt-get install ffmpeg # or sudo apt-get install vlc # Fedora/RHEL sudo dnf install mpg123 # or similar for other players # Arch sudo pacman -S mpg123 # or similar for other players

Windows/macOS: Audio playback should work out of the box.
- Check system volume
- Ensure no other audio issues on your system
- Enable debug logging withDEBUG=true
- Check the logs for any errors

- 🐍 Complete rewrite in Python for better cross-platform support
- 🔧 Improved async handling with Python's asyncio
- 📦 Simplified installation with pip
- 🛡️ Enhanced type safety with Pydantic
- 🚀 Better performance and reliability

- 💾 Added persistent state management for voice and tone preferences
- 🎯 Addedset_voicetool to set default voice
- 🎯 Addedset_tonetool to set default speaking instructions
- 🎆 Added support for new OpenAI voices: ash, ballad, and sage
- 🔄play_summarynow uses saved preferences unless overridden
- 📝 State saved to~/.summarize-mcp-state.json

- 🌍 Added Windows support using PowerShell/Windows Media Player
- 🐧 Added Linux support with auto-detection of audio players
- 🔄 Added fallback to system default audio player
- 📝 Updated documentation for multi-platform usage

- 🎉 Initial release
- ✨ Core TTS functionality with OpenAI integration
- ✨ Support for 7 different voices
- ✨ Custom speaking instructions
- ✨ Background audio playback on macOS
- ✨ Automatic file cleanup
- ✨ TypeScript implementation
- ✨ Comprehensive error handling

This tool uses OpenAI'sgpt-4o-mini-ttsmodel for text-to-speech conversion. Here's the pricing breakdown:

- 100-word summary(~30 seconds): ~$0.0075
- 500-word summary(~2.5 minutes): ~$0.0375
- 1000-word summary(~5 minutes): ~$0.075

- Length of your summaries
- Speaking speed (instructions can affect this)
- How frequently you use the tool

For current pricing details, seeOpenAI's pricing page.

This project is licensed under the MIT License - see theLICENSEfile for details.

- Built on theModel Context Protocolspecification by Anthropic
- Powered byOpenAI's TTS API
- Special thanks to the MCP community for inspiration and support

614 million contacts, live SMTP-verified. Invalid emails cost you $0. Connect Argorant with any Agent through CLI, MCP or API and build verified lists in Claude, GPT or Hermes.