MCP Play Sound Server

A Model Context Protocol (MCP) server that provides audio playback functionality for AI agents. This server enables AI agents to play sound notifications when coding tasks are completed.

⚠️ Early Version / Proof of Concept This is an early implementation currently tested only on macOS. Future versions will support additional platforms.

Features

• Audio Notifications: Play sound alerts when AI tasks complete

• Default Sound: Bundled notification sound for immediate use

• Custom Audio: Support for custom audio files (WAV, MP3, FLAC, OGG, M4A)

• Intelligent Fallback: Automatic fallback to default sound if custom audio fails

• macOS Support: Currently tested on macOS using AFPlay and SimpleAudio

Related MCP server: Audio MCP Server

Installation & Setup

Development Setup

1. Clone and install:

2. Configure Claude Desktop: Add to your claude_desktop_config.json:

3. Test: Restart Claude Desktop and ask: "Can you play a notification sound?"

Configuration

Environment variables (optional):

• CUSTOM_SOUND_PATH: Path to custom audio file

• VOLUME_LEVEL: Playback volume (0.0-1.0, default: 0.8)

• ENABLE_FALLBACK: Enable fallback to default sound (default: true)

• AUDIO_DEVICE: Specific audio output device name (e.g., "Mac Studio Speakers")

Available Tools

play_notification_sound

Plays a notification sound to alert the user.

Parameters:

• custom_sound_path (optional): Path to custom audio file

• message (optional): Context message for the notification

get_audio_status

Returns current audio system status and configuration.

test_audio_playback

Tests audio playback functionality.

Parameters:

• use_custom (optional): Test with custom sound if configured

list_audio_devices

Lists all available audio output devices on the system.

Returns:

• List of available audio devices with their properties

• Current configured device (if any)

• Default device information

Development

Prerequisites

• Python 3.10+

• uv (recommended) or pip

Testing

License

MIT License - see the LICENSE file for details.

Acknowledgments

• Model Context Protocol by Anthropic

• Python audio library maintainers