Edge TTS

You can run a lightweight, cross‑platform MCP server that converts text to speech using Microsoft Edge TTS. This server lets Claude speak with many voices and languages, without requiring external API keys. You can run it locally via Python, either from a pip install or from source, and you can configure voices, speed, volume, and pitch to fit your needs.

How to use

You start the MCP server and then connect a client that can send speak requests. Use the server to synthesize speech from text, browse and select voices, and adjust speaking rate, volume, and pitch. Typical actions include speaking a line of text, listing available voices (optionally filtered by language), and inspecting current configuration.

How to install

Prerequisites: Python 3.10+ on all platforms. No extra software is required on Windows or macOS beyond what you already have. Linux users should have ffmpeg or mpv available for audio playback.

pip install mcp-edge-tts

Additional setup (MCP configuration)

You can run the server in two common ways depending on how you installed it.

{
  "mcpServers": {
    "edge-tts": {
      "command": "python",
      "args": ["-m", "mcp_edge_tts"]
    }
  }
}

From source on Windows

If you installed from source on Windows, use the Python executable inside your virtual environment to start the MCP server.

{
  "mcpServers": {
    "edge-tts": {
      "command": "C:\\path\\to\\edge-tts-mcp\\venv\\Scripts\\python.exe",
      "args": ["-m", "mcp_edge_tts"]
    }
  }
}

From source on macOS / Linux

If you installed from source on macOS or Linux, point to the Python interpreter inside the virtual environment.

{
  "mcpServers": {
    "edge-tts": {
      "command": "/path/to/edge-tts-mcp/venv/bin/python",
      "args": ["-m", "mcp_edge_tts"]
    }
  }
}

Voice usage and configuration

While the server is running, you can issue speak requests to convert text to speech. You can also list available voices and adjust options like voice, rate, volume, and pitch to customize the output.

Configuration and environment

Configure defaults via environment variables or a .env file. You can set the default voice, speaking rate, volume, and pitch to tailor the voice output across sessions.

EDGE_TTS_VOICE=uk-UA-OstapNeural
EDGE_TTS_RATE=+0%
EDGE_TTS_VOLUME=+0%
EDGE_TTS_PITCH=+0Hz
EDGE_TTS_PLAYER=auto

Audio players

The server automatically selects an audio player suitable for your platform. On Windows you may use the built-in PowerShell MediaPlayer or alternatives like ffplay or mpv. On macOS you can use afplay or alternatives like ffplay or mpv. On Linux, ffplay is typical, with mpv as an alternative.

Popular voices

Choose from voices across languages. Examples include uk-UA-OstapNeural (Ukrainian, male), uk-UA-PolinaNeural (Ukrainian, female), en-US-AriaNeural (English US, female), en-US-GuyNeural (English US, male), de-DE-ConradNeural (German, male), fr-FR-DeniseNeural (French, female), es-ES-AlvaroNeural (Spanish, male), zh-CN-XiaoxiaoNeural (Chinese, female).

Requirements

Python 3.10 or newer is required. No extra software is needed on Windows/macOS. Linux users should ensure ffmpeg or mpv is available to handle audio playback.

Tools involved

The server exposes text-to-speech functionality through a simple API: speak, list_available_voices, and get_config. You can use these to generate speech, inspect voices, and view current settings.

Available tools

speak

Speak text aloud with optional voice, rate, volume, and pitch parameters.

list_available_voices

List all available voices, optionally filtered by language.

get_config

Show current settings and available audio players.