Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.vobiz.ai/llms.txt

Use this file to discover all available pages before exploring further.

Stream raw Vobiz audio over WebSocket into a Pipecat AI pipeline - no LiveKit or third-party WebRTC SDK required.

View on GitHub

Clone and run the full working example

​
Getting started

git clone https://github.com/vobiz-ai/Vobiz-x-Pipecat.git
cd Vobiz-x-Pipecat
pip install -r requirements.txt
python server.py

​
Overview

When a caller dials in, Vobiz is instructed via XML to open a raw WebSocket to your server. Pipecat’s WebsocketServerTransport receives the audio stream and routes it through a sequential AI pipeline - VAD → STT → LLM → TTS - before streaming synthesized audio back to the caller.

​
Architecture

Caller → Vobiz SIP → XML: <Connect><Stream url="wss://your-server/ws"/></Connect>
                                           ↓
                              WebsocketServerTransport (Pipecat)
                                           ↓
                              SileroVADAnalyzer (voice activity)
                                           ↓
                              Deepgram STT (speech → text)
                                           ↓
                              OpenAI LLM (text → response)
                                           ↓
                              ElevenLabs TTS (response → audio)
                                           ↓
                              WebSocket → Vobiz → Caller

​
How it works

1

Incoming call

When an inbound call triggers your webhook, respond with Vobiz XML instructing Vobiz to open a WebSocket stream to your server.
2

Pipecat transport

WebsocketServerTransport receives the raw audio stream. Vobiz sends base64-encoded G.711 μ-law audio frames as JSON events, which Pipecat decodes and buffers automatically.
3

Sequential pipeline

Audio flows through the Pipecat pipeline - VAD detects speech, Deepgram transcribes it, OpenAI generates a response, and ElevenLabs synthesizes speech.
4

Audio response

Encode the synthesized audio and send it back over the WebSocket. Vobiz plays it to the caller in real time.

​
Vobiz XML hook

Your webhook endpoint must return this XML when a call arrives: 
<?xml version="1.0" encoding="UTF-8"?>
<Response>
  <Connect>
    <Stream url="wss://your-server.com/ws" />
  </Connect>
</Response>

​
Environment variables

.env
DEEPGRAM_API_KEY=your-deepgram-key
OPENAI_API_KEY=sk-...
ELEVENLABS_API_KEY=your-elevenlabs-key
HTTP_PORT=8000
PUBLIC_URL=https://your-server.com  # or leave blank to use ngrok