Can Your Agent Hear You Now? Building Live Voice Agents with Gemini — Thor Schaeff
Source
youtube.com
Author
AI Engineer
Date
Why it matters
Explains that the Live API is a stateful WebSocket session with native audio rather than an STT/TTS pipeline, supporting interruptions and tool calls. Useful for deciding how to architect a real-time voice agent.
Key takeaways · AI-distilled
Schaeff chains image generations through Gemini's stateful Interactions API so characters and art style stay consistent across an illustrated story, with a fast Nano Banana image model drawing each scene.
A Live API session accepts text, audio and video and streams back audio together with a transcript. Schaeff says it supports 90 languages and can ground answers with Google Search.
Two demos show the range: audience members scanned a QR code to hear the talk in their own language through Gemini 3.5 Live Translate, and a voice radio DJ AI agentAn AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.Full definition → called Lyria 3 as a tool to generate a requested reggaeton track.
Terms in this piece · Glossary
grounding — Tying a model's answers to checkable sources — retrieved documents, live data, tool results — instead of letting it answer from memory alone.
AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
