Google has released Gemini 3.5 Live Translate, a new audio model for live speech-to-speech translation supporting 70+ languages. Unlike turn-by-turn systems, it translates continuously while preserving speaker intonation, pacing, and pitch, staying just a few seconds behind the speaker. It is available to developers via the Gemini Live API and Google AI Studio (public preview), to enterprises via Google Meet (private preview), and to all users through the Google Translate app on Android and iOS. Developer platforms like Agora, LiveKit, and Pipecat integrate with the API to simplify building voice translation apps. Google Meet gains support for 2000+ language combinations, up from English-only translation. All generated audio is watermarked with SynthID for AI content detection.
