Build Real-Time Voice Applications with Gemini 3.8 Live and 3.5 Transcribe
Google, Tuesday, September 15th, 2026
Google shows developers how to build real-time voice apps with Gemini 3.8 Live and 3.5 Transcribe.
Google walks developers through building real-time voice applications using its latest audio models, Gemini 3.8 Live, 3.8 Live Extended Thinking, and 3.5 Transcribe.
The post covers the Live API, streaming audio handling, and latency considerations for conversational interfaces.
It includes guidance on when to use transcription versus native audio dialogue. Sample patterns are provided for common voice application architectures.