Google has introduced Gemini 3.5 Live Translate, a new speech-to-speech translation model that can automatically detect more than 70 languages and generate translated audio while a person is still speaking.
The model is designed to deliver continuous translation with natural-sounding speech that preserves a speaker’s intonation, pacing, and pitch. Unlike traditional turn-based translation systems that wait for a speaker to finish before generating a response, Gemini 3.5 Live Translate processes speech as it streams, allowing conversations to continue with only a brief delay.
Google is rolling out the technology across several products. Developers can access it in public preview through the Gemini Live API and Google AI Studio, enterprise customers will begin receiving it in private preview through Google Meet, and consumers can use it through the Google Translate app on Android and iOS.
A key feature of the model is its ability to detect languages automatically without requiring users to manually select language settings. Google said the system is also designed to perform in noisy environments, making it suitable for live conversations, meetings, lessons, broadcasts, and other real-time communication scenarios.
The company is positioning the release as both a consumer product and a developer platform. Through the Gemini Live API, developers can build translation applications without managing the underlying real-time media infrastructure. Google said platforms including Agora, Fishjam, LiveKit, Pipecat, and Vision Agents have already integrated with the technology.
Several companies are testing the model in production environments. Grab is using Gemini 3.5 Live Translate to help drivers and travelers communicate during pickups, a use case that supports more than 10 million voice calls each month. Grab Chief Product Officer Philipp Kandal said the company has been evaluating the system’s ability to automatically identify multiple languages while maintaining translation accuracy and low latency.
Google is also bringing the technology to Google Meet. Speech translation in Meet will expand from five supported languages to more than 70 and support over 2,000 language combinations within a single meeting. The update is entering private preview for select Google Workspace business customers this month, with a broader rollout planned later this year.
For consumers, the model is arriving in the Google Translate app worldwide on Android and iOS. Users can connect headphones and receive translated speech that mirrors the original speaker’s tone. Google is also introducing a new listening mode for Android users, allowing translated audio to play directly through a phone’s earpiece without requiring headphones.
The launch highlights Google’s growing focus on real-time voice AI, an area where speed and natural conversation flow have become increasingly important. By generating translated speech continuously rather than waiting for complete sentences, Gemini 3.5 Live Translate aims to make multilingual conversations feel more like direct communication than a series of translated exchanges.
Google said all audio generated by the model is watermarked with SynthID, its content authentication technology designed to help identify AI-generated media.
About this article: This article was generated with AI assistance and reviewed by our editorial team to ensure it follows our editorial standards for accuracy and independence. We maintain strict fact-checking protocols and cite all sources.
Word count: 506Reading time: 0 minutes
Explore More AI Resources
Continue with high-value guides related to this topic.
