- By Prateek Levi
- Wed, 10 Jun 2026 03:38 PM (IST)
- Source:JND
Google is taking another major step in breaking down language barriers with the launch of Gemini 3.5 Live Translate, a new speech-to-speech AI translation model designed to make multilingual conversations feel far more natural.
Announced on Wednesday, the technology can listen to a speaker, translate what they're saying, and generate translated speech in near real time. Unlike traditional translation systems that often require speakers to pause and wait for a response, Google says Gemini 3.5 Live Translate works continuously, allowing conversations to flow more naturally.
The company claims the model can recognise and switch between more than 70 languages without users needing to manually select language settings. More importantly, it aims to preserve elements such as tone, pacing, and intonation, helping translated speech sound closer to the original speaker rather than a robotic voiceover.
Google says the technology is built for a wide range of scenarios, including international meetings, live broadcasts, online classes, customer support conversations, and real-time interpretation.
Designed For Real-Time Conversations
According to Google, Gemini 3.5 Live Translate processes audio while it is being spoken instead of waiting for complete sentences or conversation turns. The translated speech is generated only a few seconds behind the original speaker, reducing delays that can often make translated conversations feel awkward.
The company also says the model performs reliably in noisy or unpredictable environments, where background sounds can often affect translation accuracy.
Google Meet And Translate Get A Major Upgrade
The new translation capabilities are also making their way to Google Meet. Support is expanding from just five languages previously to more than 70 languages, significantly broadening the platform's multilingual communication capabilities.
Google is also bringing Gemini 3.5 Live Translate to the Google Translate app on both Android and iOS. Users will be able to hear translated speech through compatible connected headphones, making conversations across languages easier while on the move.
Android users are getting an additional feature called Listening Mode, which allows translated audio to play directly through the phone's earpiece.
AI Watermarking Built In
As concerns around AI-generated content continue to grow, Google says every piece of audio created by Gemini 3.5 Live Translate will include SynthID, the company's watermarking technology.
The watermark is embedded directly into generated audio and is designed to help identify content that has been produced using AI systems.
Available For Developers Too
Google is also opening access to developers through a public preview of Gemini 3.5 Live Translate in Google AI Studio and the Gemini Live API.
To make deployment easier, the company has partnered with several platforms including Agora, Fishjam, LiveKit, Pipecat, and Vision Agents. These integrations are intended to help developers quickly build voice translation experiences and multilingual applications powered by Gemini.
Google says Gemini 3.5 Live Translate is beginning its global rollout today across supported Google products and services.