Google has announced Gemini 3.5 Live Translate, its latest AI-powered speech translation model designed to enable natural, real-time multilingual communication. Built on Google’s translation technologies that process more than a trillion translated words every month across its products, the new model focuses on continuous speech-to-speech translation while preserving the original speaker’s tone, pacing, and vocal characteristics.

Unlike traditional translation systems that wait for a speaker to finish an entire sentence before generating a translation, Gemini 3.5 Live Translate operates as a streaming model. The system begins translating while the speaker is still talking, staying only a few seconds behind to maintain context and conversational flow.

Google says the technology automatically detects more than 70 languages without requiring manual language selection and can generate low-latency translated speech that sounds more natural and conversational.

One of the biggest highlights of Gemini 3.5 Live Translate is its ability to preserve vocal nuances during translation. The model maintains intonation, pacing, emphasis, and pitch while generating translated audio, helping conversations feel more natural and less robotic.

The system is also designed to function effectively in noisy and unpredictable environments, improving translation reliability during meetings, calls, public events, and travel scenarios.

Gemini 3.5 Live Translate is intended for a wide range of real-time communication use cases, including business meetings, customer support interactions, educational sessions, live broadcasts, multilingual conferences, and international collaboration. Google says the platform can also support simultaneous translation, live interpretation, and AI-powered dubbing experiences through developer integrations.

To support developers, Google is making the technology available through the Gemini Live API and Google AI Studio. Several technology partners, including Agora, Fishjam, LiveKit, Pipecat, and Vision Agents, are already integrating the model into their platforms to build multilingual communication tools, voice translation services, live interpretation systems, and real-time dubbing solutions.

Google also highlighted that ride-hailing platform Grab is testing the technology to facilitate near real-time communication between drivers and passengers across different languages. According to Google, Grab processes more than 10 million voice calls every month.

The company is also significantly expanding translation capabilities within Google Meet. Gemini 3.5 Live Translate increases language support from five languages to more than 70 languages, enabling over 2,000 language combinations within meetings. The update also introduces a redesigned interface and improvements to real-time interpretation experiences for enterprise users.

Beyond Meet, the technology is rolling out globally within the Google Translate app on Android and iOS. Users will be able to conduct real-time translated conversations while preserving the original speaker’s vocal characteristics, creating a more natural communication experience.

Google is additionally introducing a new Android Listening Mode, designed for situations where headphones are unavailable. The feature allows users to hold their smartphone to their ear like a traditional phone call while translated speech is streamed privately through the device’s earpiece speaker.

To improve transparency and reduce misuse risks, all AI-generated audio produced by Gemini 3.5 Live Translate includes SynthID watermarking, Google’s technology that embeds imperceptible identification signals into AI-generated media.

• Real-time speech-to-speech translation
• Support for 70+ languages
• Automatic language detection
• Continuous streaming translation
• Preservation of voice tone, pacing, and pitch
• Low-latency translated audio output
• Noise-resistant speech processing
• No manual language configuration required
• Real-time interpretation support
• AI-powered dubbing capabilities
• Android Listening Mode
• SynthID watermarking for AI-generated audio

• Support expanded from 5 to 70+ languages
• More than 2,000 language combinations
• Improved live interpretation experience
• Updated translation interface
• Enhanced multilingual meeting support

Gemini 3.5 Live Translate is available through:

• Gemini Live API
• Google AI Studio

Supported use cases include:

• Voice translation applications
• Live interpretation platforms
• AI dubbing solutions
• Multilingual communication services
• Enterprise collaboration tools

• Developers: Public Preview through Gemini Live API and Google AI Studio
• Google Meet: Private Preview for select Google Workspace Enterprise customers starting this month, broader rollout later this year
• Google Translate App: Rolling out globally on Android and iOS
• Android Listening Mode: Rolling out as part of the Google Translate update

Google says Gemini 3.5 Live Translate represents the next evolution of AI-powered communication, aiming to make conversations across languages feel as natural and seamless as speaking with someone in the same language.