Skip to content

ADK 的 Cartesia MCP 工具

Supported in ADKPythonTypeScript

Cartesia MCP 服务器将你的 ADK 智能体连接到 Cartesia AI 音频平台。此集成使你的智能体能够生成语音、将声音本地化为不同语言,并使用自然语言创建音频内容。

使用场景

  • Text-to-Speech Generation: Convert text into natural-sounding speech using Cartesia's diverse voice library, with control over voice selection and output format.

  • Voice Localization: Transform existing voices into different languages while preserving the original speaker's characteristics—ideal for multilingual content creation.

  • Audio Infill: Fill gaps between audio segments to create smooth transitions, useful for podcast editing or audiobook production.

  • Voice Transformation: Convert audio clips to sound like different voices from Cartesia's library.

前置条件

与智能体一起使用

from google.adk.agents import Agent
from google.adk.tools.mcp_tool import McpToolset
from google.adk.tools.mcp_tool.mcp_session_manager import StdioConnectionParams
from mcp import StdioServerParameters

CARTESIA_API_KEY = "YOUR_CARTESIA_API_KEY"

root_agent = Agent(
    model="gemini-2.5-pro",
    name="cartesia_agent",
    instruction="Help users generate speech and work with audio content",
    tools=[
        McpToolset(
            connection_params=StdioConnectionParams(
                server_params=StdioServerParameters(
                    command="uvx",
                    args=["cartesia-mcp"],
                    env={
                        "CARTESIA_API_KEY": CARTESIA_API_KEY,
                        # "OUTPUT_DIRECTORY": "/path/to/output",  # Optional
                    }
                ),
                timeout=30,
            ),
        )
    ],
)

import { LlmAgent, MCPToolset } from "@google/adk";

const CARTESIA_API_KEY = "YOUR_CARTESIA_API_KEY";

const rootAgent = new LlmAgent({
    model: "gemini-2.5-pro",
    name: "cartesia_agent",
    instruction: "Help users generate speech and work with audio content",
    tools: [
        new MCPToolset({
            type: "StdioConnectionParams",
            serverParams: {
                command: "uvx",
                args: ["cartesia-mcp"],
                env: {
                    CARTESIA_API_KEY: CARTESIA_API_KEY,
                    // OUTPUT_DIRECTORY: "/path/to/output",  // Optional
                },
            },
        }),
    ],
});

export { rootAgent };

可用工具

Tool Description
text_to_speech Convert text to audio using a specified voice
list_voices List all available Cartesia voices
get_voice Get details about a specific voice
clone_voice Clone a voice from audio samples
update_voice Update an existing voice
delete_voice Delete a voice from your library
localize_voice Transform a voice into a different language
voice_change Convert an audio file to use a different voice
infill Fill gaps between audio segments

配置

The Cartesia MCP server can be configured using environment variables:

Variable Description Required
CARTESIA_API_KEY Your Cartesia API key Yes
OUTPUT_DIRECTORY Directory to store generated audio files No

额外资源