Answering machine detection - Vobiz

Detect whether an outbound call was answered by a human or an answering machine before your AI agent starts speaking.

View on GitHub

Clone and run the full working example

Getting started

git clone https://github.com/vobiz-ai/Livekit-Vobiz-Machine-Detection-Agent-example.git
cd Livekit-Vobiz-Machine-Detection-Agent-example
pip install -r requirements.txt
python agent.py dev

Overview

When making outbound calls at scale, many calls land on voicemail. This example shows how to detect a machine answer versus a human answer, and react accordingly - leaving a pre-recorded message for voicemail, or starting a live conversation with a human.

How it works

The agent dials out via Vobiz SIP using LiveKit’s outbound calling.
When the call connects, the agent listens for audio patterns.
If short silence followed by a beep is detected → voicemail detected.
If conversational speech is detected → human answered.
The agent responds appropriately for each case.

Detection logic

Signal	Classification	Agent action
Greeting + pause + beep	Answering machine	Leave pre-recorded message
”Hello?” or conversational speech	Human	Start live conversation
Long silence	Unknown	Default to human behavior

Environment variables

.env

LIVEKIT_URL=wss://your-project.livekit.cloud
LIVEKIT_API_KEY=APIxxxxxxxxxxxxx
LIVEKIT_API_SECRET=secretxxxxxxxxxx
VOBIZ_SIP_DOMAIN=your-domain.sip.vobiz.ai
VOBIZ_USERNAME=your-username
VOBIZ_PASSWORD=your-password
OPENAI_API_KEY=sk-...
VOICEMAIL_MESSAGE_URL=https://your-server.com/voicemail.mp3