MCP Make Sound

You run a focused MCP server that adds rich audio feedback to your macOS setup. It lets your AI assistants and other MCP clients play system sounds, speak text, and play custom audio files, all controlled through simple MCP calls. This makes interactions feel more responsive and helps you stay informed without needing to look at the screen.

How to use

You connect to the MCP Make Sound server from your MCP client to trigger audio playback. Use the four available capabilities to enhance your workflows: play basic informational, warning, or error sounds; play any of the 14 built-in macOS system sounds; convert text to speech with customizable voices; and play arbitrary audio files from disk. When your AI task finishes, you can instruct the server to provide audio feedback that matches the outcome (info, warning, or error) so you can gauge progress at a glance.

How to install

Prerequisites: macOS for system sounds and afplay/say, Node.js 18 or newer, and TypeScript tooling.

1. Ensure Node.js 18+ is installed on your system.

2. Clone the project to your workstation.

git clone <repository-url>
cd mcp-make-sound

3. Install dependencies.

npm install

4. Build the project.

npm run build

5. Start the server.

npm start

Configuration and usage details

The server can be run locally as a stdio MCP server and can be configured to start automatically on launch. A complete MCP configuration example shows how to launch the server as a local process and feed its index file to Node.

Security & limitations

System sounds are limited to the 14 built-in macOS sounds. Text-to-speech supports a curated list of voices and enforces a reasonable text length. File playback requires absolute file paths and existence validation.

Example configuration (MCP)

You can configure the MCP client to launch the server locally using the following setup.

{
  "mcp-make-sound": {
    "command": "node",
    "args": [
      "/Users/nocoo/Workspace/mcp-make-sound/dist/index.js"
    ],
    "env": {},
    "working_directory": "/Users/nocoo/Workspace/mcp-make-sound",
    "start_on_launch": true
  }
}

Development

The project uses TypeScript and the MCP SDK to implement the server. It exposes four tools for sound control and runs over standard I/O. You can build, run tests, or start a development server with auto-rebuild as described in the build/run scripts.

Tools and capabilities

The server provides a set of sound-related actions you can call from your MCP client. These include simple legacy methods for quick info, warning, and error sounds, as well as a flexible play_sound method that supports system sounds, text-to-speech, and file playback.

Available tools

play_info_sound

Play an informational system sound using Glass.aiff.

play_warning_sound

Play a warning system sound using Purr.aiff.

play_error_sound

Play an error system sound using Sosumi.aiff.

play_sound

Play various sound types: system sounds, text-to-speech with optional voice, or a custom audio file from disk.