Vonage Audio Connector Server — Sample Apps

This repository contains sample applications in separate folders. Use them to explore real-time audio streaming with the Vonage Video API (Audio Connector) and Vonage Voice API, powered by Pipecat.

Repository structure

Core / foundational samples

Folder	Description
unified_sample_app	Bidirectional real-time audio over WebSocket, streaming to and from Vonage (Video API + OpenTok).
echo_server_app	Minimal WebSocket server that echoes received audio back — useful for testing the Audio Connector plumbing.

Pipecat bot samples

Folder	Description
vonage-audio-bot	Conversational STT → LLM → TTS bot supporting both Vonage Video API (Audio Connector) and Vonage Voice API (phone calls).
vonage-ac-s2s	Speech-to-speech bot using OpenAI Realtime (audio-in / audio-out) connected via Vonage Audio Connector — no separate STT/TTS step.
vonage-ac-translation-bot	Live conference translation: translates speech from one language to another in real time and injects the translated audio back into the same Vonage session.
vonage-voice-appointment-bot	Phone-based appointment booking agent: callers can check availability, book, and cancel appointments over a Vonage Voice API call.
vonage-shoe-store-audio-bot	Shoe-store voice workflow bot (Video API / Audio Connector) that captures customer details, checks inventory, creates orders, reserves a pickup window, and sends confirmation SMS via n8n webhooks.
vonage-ai-triage-nurse	AI medical triage nurse (Video API / Audio Connector) that collects patient information, checks doctor availability, schedules appointments, and sends SMS notifications — all orchestrated by n8n.
vonage-voice-mcp-server	Voice AI assistant (Voice API) that answers KB questions and manages appointments/support tickets using MCP-style HTTP tools; ships with a companion MCP tool server.

Common requirements

• Python 3.10+
• uv package manager (Pipecat samples)
• Vonage account credentials — Video Application (Application ID + Private Key) or OpenTok credentials (API key + secret), depending on the sample
• OpenAI API key (Pipecat samples)
• ngrok (or any HTTPS tunnel) for local development

Getting started

1. Pick a sample folder from the table above and open its README.md for detailed setup instructions.
2. Copy env.example to .env and fill in your credentials.
3. Start the server and point Vonage (or ngrok) at it.

Tip: Every Pipecat sample uses uv — run uv sync inside the sample folder to install dependencies into an isolated virtual environment.