Audio

The Audio MCP Server adds microphone and speaker control to your AI assistant workflows. It exposes tools for discovering available audio devices, recording from a chosen microphone, playing back the most recent recording, and playing audio files through the system speakers. This enables Claude (or other MCP-enabled assistants) to interact with your computer's audio hardware to capture user input and provide audible feedback. The server is designed to be used alongside Claude Desktop, with a hammer icon appearing in the input area once the tools are wired up.

To use it, configure Claude Desktop to point at the audio MCP server as described in the installation instructions. Once connected, you can query the available tools such as list_audio_devices, record_audio, play_latest_recording, play_audio (for TTS-like playback), and play_audio_file. Examples include listing all microphones and speakers, recording short audio clips, replaying the last recording, or playing back an MP3/ WAV file through your speakers.

Prerequisites:

• Python 3.8 or higher
• Audio input/output devices available on the system
• Access to install Python dependencies (pip)

Option A: Install via Smithery (automatic)

1. Ensure you have Node.js and npx installed.
2. Install the MCP server using Smithery:

npx -y @smithery/cli install @GongRzhe/Audio-MCP-Server --client claude

3. This will set up the server to work with Claude Desktop automatically.

Option B: Manual installation

1. Clone the repository and enter the project directory:

git clone https://github.com/GongRzhe/Audio-MCP-Server.git
cd Audio-MCP-Server

2. Create a Python virtual environment and install dependencies:

# Windows
python -m venv .venv
.venv\Scripts\activate
pip install -r requirements.txt

# macOS/Linux
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

3. Alternatively, run the included setup script to automate installation:

python setup_mcp.py

4. Start the Python audio server as appropriate for your setup (see mcp_config example).

Tips and considerations:

• Ensure your Python environment has access to your audio devices and permissions are set correctly on your OS.
• If you encounter device-not-found issues, verify the devices appear in your system's sound settings and that the correct default input/output devices are selected.
• The TTS feature is indicated as a placeholder; playback via play_audio_file or direct playback from recordings is supported with current capabilities.
• When integrating with Claude Desktop, set the paths in your Claude config to the Python interpreter and the audio_server.py script, and export PYTHONPATH as needed.
• If you see connectivity or tool discovery problems, restart Claude Desktop and double-check that the mcp_server is reachable at the expected path/port as configured.