Full Workshop: Realtime Voice AI
summary
Pipecat is an orchestration-first, open source Python framework that wires transports, STT, LLMs and TTS into real‑time voice agents; the workshop shows a runnable bot built in minutes using Pipecat open source Python framework, a Gemini Multimodal Live model to unify transcription, generation and TTS, and a local Silero VAD. Example pipeline includes dummy tools (fetch weather, restaurant recommendation), WebRTC or WebSocket transport, and phone/SIP integrations; latency targets are ~800 ms and the demo runs but conference Wi‑Fi causes flaky connectivity.
AI-written summary · about
chapters (8)
- 0:00Introduction and setup
- 10:12Voice AI fundamentals
- 17:26Pipecat pipeline overview
- 29:10Quick start and repo walkthrough
- 40:15Live coding the voice bot
- 55:33Production use and client SDKs
- 1:01:06Advanced features and demos
- 1:06:54Q&A and closing
related talks
33mDesigning Voice Agents for Real Conversations
26mMake Python Talk, Make Python Listen
23mBuilding with OpenAI: What's Ahead
51mBuilding an AI assistant that makes phone calls
25mBeyond Transcription: Building Voice AI That Understands Conversations
55mEveryday AI: Practical Applications Beyond the Chatbot
21mBuilding a Voice-Enabled AI Assistant With Javascript
19mAudio Streaming - Using WebRTC for building your own Voice AIs