Cantonese.ai MCP Server

What is Cantonese.ai MCP Server?

An MCP (Model Context Protocol) server that provides tools for text-to-speech and speech-to-text conversion using the cantonese.ai API. It is designed for developers who need to integrate Cantonese or English audio processing into MCP‑compatible clients, such as LLM agents.

How to use Cantonese.ai MCP Server?

Install with uv, set the CANTONESE_AI_API_KEY environment variable, then run uv run mcp dev server.py for development or uv run server.py for use with Claude Desktop. Two tools are exposed: text_to_speech and speech_to_text.

Key features of Cantonese.ai MCP Server

• Text‑to‑speech and speech‑to‑text via cantonese.ai API.
• Supports Cantonese and English languages.
• Secure API key handled as environment variable.
• Modern tooling with uv for fast package management.
• Easy integration with any MCP‑compatible client.

Use cases of Cantonese.ai MCP Server

• Convert Cantonese text to audio for voice applications.
• Transcribe Cantonese speech recordings into text.
• Add English TTS capabilities to an MCP‑based agent.
• Build interactive voice‑driven chatbots using MCP tools.
• Automate audio file transcription in a development workflow.

FAQ from Cantonese.ai MCP Server

What tools does the server provide?

It provides text_to_speech to convert text into an audio file and speech_to_text to transcribe an audio file into text.

Which languages are supported?

The text_to_speech tool supports both Cantonese and English. The speech_to_text tool returns the detected language in its response.

How do I obtain and set my API key?

You need an API key from cantonese.ai. Export it as the environment variable CANTONESE_AI_API_KEY (e.g., in .bashrc or .zshrc).

What are the runtime requirements?

Python 3.8 or later and the uv package manager are required. Dependencies are managed with uv sync from pyproject.toml.

How can I run the server for development or with Claude Desktop?

For development, use uv run mcp dev server.py. For integration with Claude Desktop, run uv run server.py and configure the MCP connection.