Voicevox MCP

🚀 voicevox-mcp

This project is an MCP (Model Context Protocol) server that can cooperate with the VOICEVOX engine to perform speech synthesis and obtain speaker information. It is implemented in TypeScript and utilizes the MCP SDK.

🚀 Quick Start

This project is an MCP server integrated with the VOICEVOX engine, enabling speech synthesis and speaker information retrieval. It's implemented in TypeScript with the MCP SDK.

✨ Features

• Retrieve speaker information from the VOICEVOX engine (/speakers).
• Synthesize text into speech using the specified speaker and play it locally (/speak).
  • Only supported on Mac.

📦 Installation

Start the VOICEVOX Engine (Docker recommended)

docker compose up -d

This will start the VOICEVOX engine at localhost:50021.

Install Dependencies and Build

npm install
npm run build 

💻 Usage Examples

Example Configuration for Cursor

{
  "mcpServers": {
    "voicevox-mcp": {
      "command": "node",
      "args": ["${Path to Repository}/dist/index.js"],
      "env": {
        "SPEAKER_ID": 8,
        "SPEED_SCALE": 1.2,
        "VOICEVOX_API_URL": "http://localhost:50021" 
      }
    }
  }
}

Set the VOICEVOX_API_URL as needed.

• You can get the list of speakers from the MCP client using the speakers tool.
• You can synthesize text into speech and play it locally using the speak tool (Recommended on Mac as it uses the afplay command).

Main Dependencies

• @modelcontextprotocol/sdk
• zod
• typescript

📚 Documentation

Important Notes

• Future Improvements
  • Speech synthesis won't work if the VOICEVOX engine isn't running at localhost:50021.
  • On non-Mac environments, modify the afplay part accordingly.

📄 License

MIT License