The Zonos TTS MCP Server enables text-to-speech functionality in Claude through the speak_response tool, allowing it to generate and play spoken audio from text with the following capabilities:
Multi-Language Support: Generate speech in different languages (default: en-us)
Emotion Control: Customize the emotional tone (neutral, happy, sad, angry)
PulseAudio Integration: Ensures proper audio playback
MCP Integration: Seamlessly works with Claude's Model Context Protocol
The server requires a running instance of Zonos API with proper Node.js and PulseAudio setup.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@Zonos TTS MCP Server speak this in French with a happy tone: Bonjour, comment allez-vous?"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
A Model Context Protocol integration for Zonos TTS, allowing Claude to generate speech directly.
To install Zonos TTS Integration for Claude Desktop automatically via Smithery:
npx -y @smithery/cli install @PhialsBasement/zonos-tts-mcp --client claudeMake sure you have Zonos running with our API implementation (PhialsBasement/zonos-api)
Install dependencies:
npm install @modelcontextprotocol/sdk axiosConfigure PulseAudio access:
# Your pulse audio should be properly configured for audio playback
# The MCP server will automatically try to connect to your pulse serverBuild the MCP server:
npm run build
# This will create the dist folder with the compiled serverAdd to Claude's config file:
Edit your Claude config file (usually in ~/.config/claude/config.json) and add this to the mcpServers section:
"zonos-tts": {
"command": "node",
"args": [
"/path/to/your/zonos-mcp/dist/server.js"
]
}Replace /path/to/your/zonos-mcp with the actual path where you installed the MCP server.
Related MCP server: ElevenLabs MCP Server
Once configured, Claude automatically knows how to use the speak_response tool:
speak_response(
text="Your text here",
language="en-us", # optional, defaults to en-us
emotion="happy" # optional: "neutral", "happy", "sad", "angry"
)Text-to-speech through Claude
Multiple emotions support
Multi-language support
Proper audio playback through PulseAudio
Node.js
PulseAudio setup
Running instance of Zonos API (PhialsBasement/zonos-api)
Working audio output device
Make sure both the Zonos API server and this MCP server are running
Audio playback requires proper PulseAudio configuration
Inputs: text (required), language, emotion.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 1 tool
With only one tool, there is no possibility of ambiguity or overlap between tools. The tool 'speak_response' stands alone with no other tools to confuse it with, making disambiguation perfect.
Since there is only one tool, naming consistency is inherently perfect. There are no other tools to compare against, so no inconsistencies can exist in the naming pattern.
A single tool for a TTS (Text-to-Speech) server feels thin and under-scoped. TTS typically involves operations like generating, streaming, or managing speech, but this server only offers one tool, which is insufficient for a complete TTS domain coverage.
The server is severely incomplete for a TTS domain. With only a 'speak_response' tool and no description, it lacks essential operations such as configuring voices, controlling speech parameters, or handling different input formats, making it inadequate for typical TTS workflows.
Claude makes real phone calls for you — in many languages, with transcript and outcome back in chat.
Intimate voice AI: 29 voices, 22 languages, emoji vocal cues, multi-voice scenes.
Transcribe audio and video with Speechmatics speech-to-text from Claude and any MCP client.
Speech, transcription, voice agents, Trace, Recap, dubbing and narration with browser OAuth.