@senchabrew/gemini-tts-mcp

An MCP (Model Context Protocol) server that wraps Google Gemini TTS (gemini-2.5-flash-preview-tts) for use with Claude Code, Claude Desktop, Cowork, or any MCP-compliant client.

Generate natural-sounding speech audio from text, with 30+ prebuilt voices and natural-language direction prompts (e.g. "in a frustrated tone", "slowly and clearly").

Installation

Add the server to your client's MCP configuration. For Claude Desktop, edit ~/Library/Application Support/Claude/claude_desktop_config.json (macOS):

{
  "mcpServers": {
    "voice": {
      "command": "npx",
      "args": ["-y", "@senchabrew/gemini-tts-mcp"],
      "env": {
        "GEMINI_API_KEY": "AIza..."
      }
    }
  }
}

Get a Gemini API key at https://aistudio.google.com/apikey. Restart your client to load the server.

Requires Node.js 20 or later.

Tools

generate-voice

Generate speech audio and save it as a WAV file (24 kHz, mono, 16-bit PCM).

Input:

| Field | Type | Required | Description | |---|---|---|---| | prompt | string | yes | The full TTS prompt. Embed direction in the prefix, then the speech text in quotes — e.g. 「明るい声で：「こんにちは、今日もよろしくお願いします」」 | | voice_name | string | yes | One of the 30 prebuilt voices (see list-voices). | | output_path | string | yes | Absolute path where the WAV file will be written. Parent directories are auto-created. |

Output:

{ "success": true, "output_path": "/path/to/out.wav", "duration_ms": 3175 }

On HTTP 429, the server retries up to 3 times with a 60-second backoff before failing.

list-voices

Returns the array of available voice names.

Input: none.

Output:

{ "voices": ["Zephyr", "Puck", "Charon", "Kore", ...] }

Available voices

Zephyr, Puck, Charon, Kore, Fenrir, Leda, Orus, Aoede,
Callirrhoe, Autonoe, Enceladus, Iapetus, Umbriel, Algieba,
Despina, Erinome, Algenib, Rasalgethi, Laomedeia, Achernar,
Alnilam, Schedar, Gacrux, Pulcherrima, Achird, Zubenelgenubi,
Vindemiatrix, Sadachbia, Sadaltager, Sulafat

Refer to the Gemini TTS docs for voice character previews.

Rate limits

Google's Gemini API rate limits as of 2026:

| Tier | RPM | RPD | |---|---|---| | Free | 10 | 250 | | Paid Tier 1 | 300 | 10,000 |

For Free tier, pace calls at ~7 seconds apart from the calling side. The MCP server itself does not throttle between calls — only retries on 429.

License

MIT