1. Pipecat Explained: Building Real-Time Voice AI (WebRTC + Code Example)
📘 Pipecat + WebRTC Voice AI — Full Context Guide
00:00 Introduction
00:56 What Is a Voice AI Agent?
02:00 Why Real-Time Is Non-Negotiable
02:44 What Is WebRTC (aiortc)?
03:56 What Is Pipecat?
04:29 High-Level Architecture
06:02 Frames (Pipecat’s Core Abstraction)
07:03 What Is a Processor?
08:22 What a Processor REALLY Is
09:40 VAD (Voice Activity Detection)
11:12 Turn Analyzer
12:40 Why VAD & Turn-Taking Live in Transport
14:06 The Pipecat Pipeline Concept
15:18 Pipeline Order (Why It Matters)
16:23 Pipeline vs Task vs Runner
17:33 Minimal Pipecat Code Example
18:13 RTVI (Real-time Voice Intelligence)
19:18 Observers in Pipecat
20:19 RTVI vs RTVIObserver
22:03 Do Observers Get Notified at Every Step?
22:54 Metrics in Pipecat
23:30 What Happens When a User Connects for First time?
24:33 Pipecat in One Mental Picture
25:22 Putting it all together
25:58 Summary
26:19 Code Walkthrough
Slides: https://docs.google.com/presentation/...
In this video, we break down how real-time Voice AI agents actually work, using Pipecat + WebRTC (aiortc) as the core example.