Gemini 3.5 Live Translate Brings Near Real-Time Speech Translation to Google AI Studio, Meet, and Translate
Google's new Gemini 3.5 Live Translate model translates speech in over 70 languages nearly in real time across AI Studio, Meet, and Translate.
What Google announced
Google introduced Gemini 3.5 Live Translate, describing it as its latest audio model for near real-time speech-to-speech translation in over 70 languages. According to a post on the Google blog, written by a Google senior staff software engineer, the model automatically detects 70+ languages and generates smooth, natural-sounding translated speech that preserves a speaker's intonation, pacing, and pitch. Google frames the release against its own translation history, noting that translation at Google began 20 years ago as one of the company's pioneering machine learning experiments, and that Google now translates over a trillion words every month for billions of users across its products.
How the translation works
Google contrasts the model with turn-by-turn systems that wait for a speaker to finish before responding. The company says 3.5 Live Translate instead generates speech continuously, balancing the trade-off between waiting for more context to improve quality and translating right away to stay in sync with the speaker. Google says this produces fluid audio without awkward pauses, and that the model stays only a few seconds behind the speaker throughout a session. The post also says the model processes speech as it streams in, handles multilingual input without manual configuration, and includes noise robustness meant to handle loud, unpredictable environments such as live calls, meetings, lessons, and broadcasts.
Where it is rolling out
Google says Gemini 3.5 Live Translate is rolling out starting today: for developers in public preview through the Gemini Live API and Google AI Studio, for enterprises in private preview this month in Google Meet, and for everyone through the Google Translate app on Android and iOS. Developer platforms including Agora, Fishjam, LiveKit, Pipecat, and Vision Agents already use the Gemini Live API to build and deploy voice translation apps, according to the post. Google Meet's speech translation will move from five languages and English-only pairs to more than 70 languages and over 2000 language combinations, and Google says the Meet rollout starts in private preview for select business Google Workspace customers this month, followed by a broader rollout later this year.
Early users and safety
Google says its partner Grab is testing the model to help drivers and travelers communicate across languages at pickups, noting that Grab's users make over 10 million voice calls per month through the platform. Google also cites positive early feedback from companies including CJ ENM and LiveKit on the model's translation quality, accuracy, and low latency. Every audio clip the models generate carries an imperceptible SynthID watermark, which Google says is meant to keep AI-generated audio detectable and help prevent misinformation.
Key facts
- Gemini 3.5 Live Translate automatically detects speech in over 70 languages and translates it in near real time.
- Google says the model stays only a few seconds behind the speaker and avoids the pauses of turn-by-turn systems.
- Rollout starts today: public preview for developers via the Gemini Live API and AI Studio, private preview this month in Google Meet, and availability in the Google Translate app on Android and iOS.
- Meet's speech translation moves from five languages and English-only pairs to more than 70 languages and over 2000 language combinations.
- Partner Grab, whose users make over 10 million voice calls per month, is testing the model for driver-traveler communication.
Why it matters
Teams building calls, meetings, lessons, or broadcasts that cross languages can now reach for a translation model Google says holds pace with a live speaker instead of waiting for pauses, through the same Gemini Live API that platforms such as LiveKit and Pipecat are already wiring into voice apps.
Check your understanding
How many languages does Google say Gemini 3.5 Live Translate automatically detects?
Which three Google products is Gemini 3.5 Live Translate rolling out to first?