Skip to main content
Glama
ross-sec

Kokoro TTS MCP Server

by ross-sec
OverviewSchemaSourceRelated ServersDiscussionsAdmin
JavaScript
Local

Ross Technologies Logo

MCP server for text-to-speech using Kokoro TTS

npm version License: MIT GitHub

100% Local • No Python Required • Female Voice Default

📦 NPM Package | 🐙 GitHub Repository | 📚 Documentation

Overview

A production-ready MCP server that provides text-to-speech capabilities using the Kokoro TTS model. Features a default female voice (af_heart) and runs 100% locally using native JavaScript, eliminating Python dependencies.

Built by Ross Technologies
📍 Beer Sheva, Israel | 📧 devops.ross@gmail.com

Related MCP server: voice-audio-mcp

Features

✅ 100% Local - No external API calls, complete privacy
✅ Native JavaScript - Built with TypeScript and Node.js
✅ SSE & Stdio Support - Multiple transport modes
✅ NPX Ready - Run directly without installation
✅ Female Voice Default - Uses af_heart voice out of the box
✅ Auto Audio Playback - Automatically saves and plays generated audio

Quick Start

# Run with npx (no installation)
npx @ross_tchnologies/kokoro-tts-mcp-server

# Or install globally
npm install -g @ross_tchnologies/kokoro-tts-mcp-server
kokoro-tts-mcp

Installation

NPM

npm install @ross_tchnologies/kokoro-tts-mcp-server

NPX (No Installation)

npx @ross_tchnologies/kokoro-tts-mcp-server

Global Installation

npm install -g @ross_tchnologies/kokoro-tts-mcp-server

Visit Installation Guide for detailed instructions.

MCP Client Configuration

Add to your MCP configuration file (e.g., ~/.cursor/mcp.json):

{
  "mcpServers": {
    "kokoro-tts": {
      "command": "npx",
      "args": ["-y", "@ross_tchnologies/kokoro-tts-mcp-server"]
    }
  }
}

See Configuration Guide for more options.

Usage

Text to Speech Tool

{
  "name": "text_to_speech",
  "arguments": {
    "text": "Hello, world!",
    "voice": "af_heart",  // optional, default
    "speed": 1.0          // optional, default
  }
}

Available Voices: af_heart (default), af_bella, af_sarah, and more.

See API Reference for complete documentation.

Documentation

📚 Full Documentation Wiki

Quick Links:

Requirements

  • Node.js v18 or higher

  • No Python or other external dependencies required! 🎉

Troubleshooting

  • First run: Downloads ~300MB model (one-time, takes 2-5 minutes)

  • Audio not playing: File is saved - check response message for location

  • WSL users: Copy audio files to Windows Desktop for playback

See Troubleshooting Guide for detailed solutions.

Contributing

Contributions are welcome! See:

Links & Resources

License

MIT License - See LICENSE for details.

Copyright © 2025 Ross Technologies. All rights reserved.

Made with ❤️ by Ross Technologies

Back to Top

This server cannot be deployed

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Maintenance

ActivityInactive
ResponsivenessSyncing

Resources

Related MCP Connectors

Related MCP Servers