How to Build Your First Real Time Voice Agent in Python...
summary
An English tutor called “English Speaking Buddy” is built in Python using AWS services and Pipecat and it actually runs: real-time audio → text → LLM → audio pipeline with WebRTC capture, VAD, ASR, LLM, and TTS, plus short‑ and long‑term memory backed by a vector store. The presentation gives concrete fixes for the usual failures—measureable latency per turn, pick an efficient model, use prompt caching, add logging/observability and API‑key handling—and shows pronunciation/grammar detection and a deployed prototype.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 2:57Architectural approaches
- 5:45Building voice agents
- 8:59Live demo
- 15:40Prototype to production
- 21:22Deploying with AWS
- 24:33Q&A
related talks
33mDesigning Voice Agents for Real Conversations
01h 58mWhat If Your Backend Could Talk? Building Real-Time AI Voice Agents
17mBuilding Effective Voice Agents
32mBuilding LLM Voice Bots with Open Source Tools
01h 02m[Full Workshop] Building Conversational AI Agents
01h 26mBuilding voice agents with OpenAI
20m10 Things I Learned Building Voice Agents in 2025
29mAndroid Gets a Voice: Building Real-Time Agents