minimax-tts-mcp

A lightweight MCP (Model Context Protocol) server that gives AI the ability to speak. Text in, audio URL out.

Built with Flask + MiniMax TTS API.

What it does

• Exposes a generate_speech tool via MCP protocol

• AI sends text → server calls MiniMax TTS → returns a playable mp3 URL

• Supports emotion control, speed adjustment, and language selection (Cantonese, Mandarin, etc.)

Quick Start

1. Clone & install

git clone https://github.com/YOUR_USERNAME/minimax-tts-mcp.git
cd minimax-tts-mcp
pip install -r requirements.txt

2. Configure

cp .env.example .env
# Edit .env with your MiniMax API key and voice ID

Get your API key and voice ID from MiniMax Platform.

3. Run

python server.py

Server starts at http://localhost:5000.

Endpoints

MCP Integration

Add to your MCP client config:

{
  "mcpServers": {
    "tts": {
      "url": "http://localhost:5000/mcp"
    }
  }
}

The server exposes one tool:

generate_speech

Direct TTS Usage

curl -X POST http://localhost:5000/tts \
  -H "Content-Type: application/json" \
  -d '{"text": "Hello world", "speed": 1.0}' \
  --output speech.mp3

Deploy

Works anywhere that runs Python — a VPS, cloud VM, or container. For public access, put it behind a reverse proxy (nginx/caddy) with HTTPS.

Example with systemd:

# /etc/systemd/system/tts-mcp.service
[Unit]
Description=MiniMax TTS MCP Server

[Service]
WorkingDirectory=/path/to/minimax-tts-mcp
EnvironmentFile=/path/to/minimax-tts-mcp/.env
ExecStart=/usr/bin/python3 server.py
Restart=always

[Install]
WantedBy=multi-user.target

Notes

• Generated audio files are stored in /tmp and not auto-cleaned. For production, add a cron job or cleanup logic.

• MiniMax API is a paid service. Check their pricing for TTS usage.

• Voice cloning requires setting up a custom voice on MiniMax platform first.

License

MIT