TTS MCP Server

A Model Context Protocol (MCP) server that provides Text-to-Speech (TTS) functionality using Google TTS API.

Features

• Tools: Generate TTS audio, list voices, get style templates

• Resources: Access voice style templates via URI patterns

• Prompts: Direct access to 16+ voice style prompt templates

Related MCP server: Edge-TTS MCP Server

Installation

1. Clone this repository

2. Install dependencies:

   npm install

3. Build the project:

   npm run build

Usage

Run the server:

npm start

Or for development:

npm run dev

MCP Configuration

Add the following to your MCP configuration file (e.g., mcp.json):

{
  "mcpServers": {
    "tts-mcp-server": {
      "command": "npx",
      "args": [
        "-y",
        "@activebook/tts-mcp-server"
      ],
      "env": {
        "GOOGLE_API_KEY": "{your-google-api-key}", // required, change to your own key
        "GOOGLE_NAME_MODEL": "gemini-2.0-flash", // optional
        "GOOGLE_TTS_MODEL": "gemini-2.5-flash-preview-tts", // optional
        "GOOGLE_VOICE": "kore" // optional
      }
    }
  }
}

Tools

google_tts_generate

Description: Generate TTS audio from text content

Parameters:

• content: string (The text content to convert to speech) [required]

• directory: string (Directory to save the audio file (Default: current directory)) [optional]

• voice: string (Voice to use for TTS (Default: Kore)) [optional]

• style: string (Speech style - either a template name like 'news_tone' or custom style text) [optional]

Returns: Path to the saved audio file

get_google_tts_voices

Description: Get list of available TTS voices

Parameters:

• count: number (Number of voices to return (0 for all)) [optional]

get_voice_styles

Description: Get list of predefined speech style templates

Parameters:

• detail: boolean (Whether to show voice style detail (Default: false)) [optional] Returns: JSON object containing available speech style templates with descriptions and style prompts

Prompts

Access voice style prompt templates directly:

Available Prompts

• news_anchor - Formal, clear news-reporting style

• professional_presenter - Confident business presentation style

• bedtime_storyteller - Gentle, soothing storytelling voice

• motivational_speaker - Energetic, inspiring delivery

• podcast_host - Engaging conversational style

• documentary_narrator - Calm, authoritative narration

• mysterious_narrator - Deep, suspenseful voice

• humorous_comedian - Light-hearted, comedic tone

• calm_meditation_guide - Soothing meditation voice

• dramatic_performer - Expressive theatrical style

• And 6 more styles...

Usage

{
  "method": "prompts/get",
  "params": {
    "name": "news_anchor"
  }
}

Returns: Style prompt text ready for TTS generation

Resources

Access voice style metadata via URI patterns:

Voice Style Templates

• URI Pattern: tts://voice-styles/{style_name}

• Example: tts://voice-styles/news_anchor

• Content: JSON with style description and prompt

Usage

{
  "method": "resources/read",
  "params": {
    "uri": "tts://voice-styles/bedtime_storyteller"
  }
}

Returns: Voice style metadata in JSON format

Screenshots