LiveKit Vobiz Outbound Agent 📞

A production-ready voice agent capable of making outbound calls using LiveKit, Deepgram, and Groq (Llama 3.3).
Designed for reliability, speed, and ease of deployment.

🚀 Features

• Ultra-Fast LLM: Uses Groq running llama-3.3-70b-versatile for near-instant responses.
• High-Quality Audio: Uses Deepgram for both Speech-to-Text (STT) and Text-to-Speech (TTS).
• SIP Trunking: Integrated with Vobiz for PSTN connectivity.
• Robust Configuration: Centralized config.py for easy customization of prompts, models, and voices.

---

🛠️ Setup & Installation

1. Prerequisites

• Python 3.10+ (Recommended: 3.10.13)
• A LiveKit Cloud account
• A Deepgram API Key
• A Groq API Key
• A SIP Provider (e.g., Vobiz)

2. Clone & Install

# Clone the repository
git clone <your-repo-url>
cd basic-conversation-agent

# Create a virtual environment
python3 -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

3. Configure Environment

Copy the example environment file and fill in your credentials:

cp .env.example .env
nano .env  # Or open in your editor

Required Variables:

• LIVEKIT_URL, LIVEKIT_API_KEY, LIVEKIT_SECRET
• DEEPGRAM_API_KEY
• GROQ_API_KEY
• VOBIZ_SIP_* variables (for outbound calls)

---

🏃‍♂️ Usage

1. Start the Agent

This runs the agent process which listens for room connections.

python agent.py start

2. Make an Outbound Call

In a new terminal window (ensure venv is active), run:

python make_call.py --to +91XXXXXXXXXX

Note: The number must include the country code (e.g., +1 or +91).

---

🔧 Troubleshooting Guide

❌ Error: model_decommissioned (Groq/Llama)

Cause: The configured LLM model is no longer supported by Groq.
Fix:

1. Open config.py.
2. Update GROQ_MODEL to a supported model (e.g., llama-3.3-70b-versatile or llama-3.1-8b-instant).
3. Restart agent.py to apply changes.

❌ Error: 404 Not Found (SIP Trunk)

Cause: The SIP_TRUNK_ID in .env is incorrect or doesn't exist in your LiveKit project.
Fix:

1. Run python list_trunks.py to see available trunks.
2. If none exist, run python create_trunk.py to create one.
3. Update .env with the correct ID.

❌ Error: Address already in use (Port 8081)

Cause: Another instance of agent.py is already running.
Fix:

1. Find the process: lsof -i :8081
2. Kill it: kill -9 <PID> or pkill -f "python agent.py"

❌ Error: No module named 'certifi' or other imports

Cause: Dependencies are missing.
Fix:

1. Ensure your virtual environment is active (source venv/bin/activate).
2. Run pip install -r requirements.txt.

❌ Call Connects but No Audio

Cause: TTS (Text-to-Speech) failure or WebSocket issues.
Fix:

1. Check terminal logs for APIStatusError.
2. If using OpenAI TTS, ensure you have OpenAI credits.
3. Recommended: Switch to Deepgram TTS (set TTS_PROVIDER=deepgram in .env).

---

📂 Project Structure

• agent.py: Main application logic.
• config.py: Central configuration for prompts, models, and constants.
• make_call.py: Script to initiate outbound calls.
• create_trunk.py / setup_trunk.py: Utilities for SIP trunk management.

LIvekitAIVoice