Google has introduced Gemini 3.5 Live Translate, a new audio model designed for real-time speech-to-speech translation. The system builds on two decades of machine learning work in translation and is aimed at enabling more natural, continuous communication across languages while preserving tone, pacing, and pitch.
Google reports that its translation systems now process over a trillion words each month across its products, and this update represents the next step in improving live multilingual interaction.
Gemini 3.5 Live Translate: Real-time speech translation with natural voice flow
Gemini 3.5 Live Translate is built for continuous streaming translation instead of traditional turn-based systems. It generates translated speech while the speaker is still talking, staying only a few seconds behind to maintain alignment.
Key capabilities include:
- Automatic detection of 70+ languages
- Continuous speech-to-speech translation without waiting for sentence completion
- Preservation of intonation, pacing, and pitch
- Balance between waiting for context and immediate translation to improve quality and speed
- Low-latency output with smooth audio and no awkward pauses
- Strong noise robustness for unpredictable environments
- No manual language configuration required
Multilingual communication and live use cases
The system is designed for real-time communication scenarios such as meetings, calls, lessons, and broadcasts. It processes speech as it streams, enabling smoother multilingual conversations and supporting live interpretation.
It also enables use cases like simultaneous multi-language translation and real-time dubbing through developer integrations.
Developer ecosystem and integrations
Through the Gemini Live API and Google AI Studio, developers can build real-time voice translation applications while relying on Google’s streaming infrastructure.
Platforms including Agora, Fishjam, LiveKit, Pipecat, and Vision Agents are integrating the model to support:
- Voice translation applications
- Live interpretation systems
- Real-time dubbing tools
- Multilingual communication services
Google also notes that Grab is testing the model to support near real-time communication between drivers and passengers. The platform handles over 10 million voice calls per month.
Google Meet integration
In Google Meet, Gemini 3.5 Live Translate will improve multilingual meetings by expanding real-time translation capabilities.
Updates include:
- Support for 70+ languages, up from five previously
- More than 2,000 language combinations in a single meeting (previously limited to English-based translation)
- Updated interface for faster access to speech translation tools
- Improved real-time interpretation experience for meetings
The feature is rolling out in private preview for select Google Workspace enterprise customers starting this month, followed by a broader rollout later this year.
Google Translate app experience
The update is also rolling out globally in the Google Translate app on Android and iOS.
Key features include:
- Live translation using headphones with natural tone preservation
- Support for 70+ languages in real-time conversation mode
- More seamless speech output that mirrors the speaker’s tone
Android listening mode
A new Android listening mode is also being introduced:
- Users can hold the phone to their ear like a call
- Translated audio streams directly through the earpiece
- Designed for situations without headphones or for private listening
Safety and watermarking
All audio generated by Gemini 3.5 Live Translate includes SynthID watermarking, which embeds an imperceptible signal into the output. This allows AI-generated audio to be detected and helps improve transparency and reduce misuse risks.
Availability and rollout
Gemini 3.5 Live Translate is rolling out across Google products in phases:
- Developers: Public preview via Gemini Live API and Google AI Studio
- Google Meet: Private preview for select Workspace enterprise customers starting this month, with broader rollout planned later this year
- Google Translate app: Global rollout on Android and iOS
- Android: Listening mode for earpiece-based translation experience