Voice Call (Twilio)

🚀 Voice Call MCP Server

This is a real-time voice call server project built with Twilio and OpenAI. The server supports multi-language calls, can handle complex phone business logic, and provides detailed call records and analysis.

🚀 Quick Start

Prerequisites

• System Requirements:
  • Operating System: Linux (recommended) or macOS
  • Memory: At least 4GB
  • CPU: Dual-core or higher
• Dependencies:
  1. Node.js (>=16.0)
  2. PostgreSQL or other supported databases
  3. Twilio account
  4. OpenAI API key

Installation Steps

# Clone the repository
git clone https://github.com/your-repo.git
cd your-repo

# Install dependencies
npm install

# Configure environment variables
export TWILIO_ACCOUNT_SID=your_sid
export TWILIO_AUTH_TOKEN=your_token
export OPENAI_API_KEY=your_key

Configuration

Environment Variables

Create a .env file in the project root directory and add the following content:

TWILIO_ACCOUNT_SID=your_account SID
TWILIO_AUTH_TOKEN=your_auth token
OPENAI_API_KEY=your_openai key
DB_URI=postgresql://localhost:5432/voip_calls

Business Logic

Edit the config/businessLogic.js file and add custom call rules:

module.exports = {
    languageSupport: ['zh-CN', 'en-US', 'de-DE'],
    fallbackLanguage: 'zh-CN',
    callTimeout: 30000 // 30 seconds timeout
};

✨ Features

• Multi-language Support: Supports real-time voice recognition and generation in multiple languages.
• Intelligent Routing: Automatically routes calls to different processing nodes according to rules.
• Recording Function: Saves audio files of all calls.
• Data Analysis: Provides detailed analysis reports such as call duration and user intent.

📦 Sequence Diagram

Call Flow

1. User Initiates Call: Initiates a voice call through Twilio.
2. AI Answers: OpenAI's real-time voice model starts processing the call content.
3. Intelligent Interaction: The AI converses with the user based on the preset business logic.
4. Recording and Archiving: All call content is automatically stored in the local database.

💻 Usage Examples

Intelligent Customer Service

// Customer service logic configuration
const { createVoiceCall } = require('voip-service');

createVoiceCall({
    fromNumber: '1234567890',
    toNumber: '0987654321',
    language: 'zh-CN',
    callbackUrl: 'http://localhost:3000/api/call-back'
}).then(call => {
    console.log('Call created:', call.sid);
});

🔧 Why Choose MCP

• High Performance: Based on Twilio's reliable communication platform and OpenAI's powerful AI capabilities.
• Easy to Expand: Supports quickly adding new languages and business logic.
• High Availability: Built-in fault tolerance mechanism ensures stable service operation.

⚠️ Important Note

1. Phone Number Format: All phone numbers must use the E.164 standard (e.g., +1234567890).
2. Rate Limits: Pay attention to the rate limits and billing of your Twilio and OpenAI accounts.
3. Voice Conversation: The AI processes natural conversations in real-time.
4. Call Duration: Control the call duration to save costs.
5. Public Exposure: The random URL of the ngrok tunnel is publicly exposed, so pay attention to security.

🛠️ Troubleshooting

Common Issues

1. "Phone number must be in E.164 format"
   • Ensure that the number starts with a "+" and includes the country code.
2. "Invalid credentials"
   • Check if TWILIO_ACCOUNT_SID and TWILIO_AUTH_TOKEN are correct.
3. "OpenAI API error"
   • Verify that OPENAI_API_KEY is valid and has sufficient credits.
4. "Failed to start ngrok tunnel"
   • Ensure that NGROK_AUTHTOKEN is valid and has not expired.

🤝 Contribution Guidelines

Contributions are welcome! Here are the areas to be improved:

• Support more AI models.
• Add database integration to store conversation history.
• Improve latency and response time to enhance the call experience.
• Strengthen error handling and recovery mechanisms.
• Add more preset conversation templates.

📄 License

This project is licensed under the MIT License. For details, please refer to the LICENSE file.

For more information, please visit our Documentation Page or join our Developer Community.