Google says its Gemini 3.5 Live Translate model will arrive in Google Meet later this year, bringing speech-to-speech translation that can detect more than 70 languages and support 2,000+ language combinations in a single meeting. The rollout begins as a priva
By the time a meeting ends, the conversation usually ends too — at least when languages get in the way. Google is trying to change that timing. The company has announced Gemini 3.5 Live Translate. a live audio speech-to-speech translation model meant to make real-time conversations feel less like interpreting and more like speaking.
Google says the model can automatically detect over 70 languages. It’s also designed to keep the speaker’s intonation. pacing. and pitch. so translated speech doesn’t just sound correct — it’s meant to sound natural. Instead of waiting for someone to finish. the system continuously generates translated speech. in contrast to turn-by-turn approaches that pause translation until the speaker stops.
Google isn’t treating this as a niche experiment. The announcement points to demos showing the model in action. including a live dubbing scenario where a video is translated into another language in real time. and a separate demo focused on translating talks and presentations across multiple languages.
The biggest change comes with Google Meet. Gemini 3.5 Live Translate will power speech translation in Meet later this year. and Google says that will raise language support from the current limit of five languages to 70-plus. In the same meeting. it can enable over 2. 000 language combinations — a shift from Meet’s prior speech translation that only translated to and from English.
Google also plans to update Meet’s UI so speech translation is easier to access faster. When people are trying to keep up during a call, the difference between “available” and “easy to reach” can matter.
The rollout begins with a careful first step. The improved Meet speech translation is scheduled to initially roll out as a private preview for select business Google Workspace customers later this month. with the timing left unspecified. After that initial group, Google says it will expand the feature to more users later this year.
Outside of Meet, Gemini 3.5 Live Translate is also coming to the Google Translate app globally on Android and iOS. With the release, Google is introducing a new “listening mode” for Android. That mode is designed for headphones. letting users listen to live translations. and it also adds the ability to hear translations through the phone’s earpiece.
Because this is AI-generated audio, Google is also building in a layer of provenance. The audio produced by Gemini 3.5 Live Translate will be watermarked with SynthID. Google says this is an invisible watermark woven directly into the audio output to help prevent misinformation.
Google Gemini 3.5 Live Translate Google Meet translation speech-to-speech translation Google Translate SynthID watermark Android listening mode real-time translation Google Workspace private preview
So it translates 70 languages now? Cool but can it translate my roommates correctly too?
I swear every time they say “natural” translation it still sounds robotic. Like the pitch thing… okay but will it mess up names and dates? Also “later this year” like when.
Wait, so Meet is raising from 5 to 70? I thought Meet already did like Spanish and French automatically. Maybe I’m using it wrong or they’re talking about something different.
This is gonna be great until it hears accents and decides to “detect” the wrong language lol. The part about not waiting for someone to finish sounds like it could interrupt you mid-sentence. Also 2,000+ combinations sounds made up like tech always does.