At Google I/O 2025, the Mountain View giant unveiled a new Gemini-powered speech translation feature for Google Meet that promises to break down language barriers in real time, without losing the speaker’s voice, tone, or expression.
Google Meet’s real-time AI speech translation begins rolling out in beta today for Google AI Pro and Ultra subscribers. It uses generative AI to translate spoken language during video calls and overlay it with a natural-sounding version of the speaker’s voice in the listener’s preferred language. The initial rollout supports English and Spanish, with Italian, German, and Portuguese slated to arrive in the coming weeks.
In a live demo during I/O 2025, Google showcased how the technology works: an English-speaking user on a Meet call with a Spanish-speaking colleague was able to converse seamlessly. Once the Gemini translation was activated, the colleague’s Spanish was dubbed into English, not with a generic robotic voice, but one that closely matched the speaker’s natural cadence and intonation. The reverse worked just as smoothly.
Unlike older machine translation tools, which often flatten speech into a monotone voice, Google’s new feature aims to make multilingual conversations feel as authentic as monolingual ones. The translated speech captures vocal inflexions and expressions, adding nuance and personality to what’s being said, which is a crucial step in making cross-lingual collaboration feel natural, not mechanical.
While competitors like Skype and Microsoft Teams have offered real-time translation in the past, Google’s approach may finally deliver the kind of fluid experience businesses and global teams have long hoped for.
The speech translation feature is currently exclusive to paid users but will expand to Google Workspace and enterprise customers later this year. Google hasn’t revealed a timeline for when free-tier users might gain access.
Also Read: Gmail’s AI smart replies can now respond just like you!