Real-Time Speech Translation in Google Meet

At Google I/O 2025, Google introduced a remarkable new feature for Google Meet: real-time speech translation. This advanced capability, powered by a large language audio model created by Google DeepMind, allows for seamless multilingual dialogues by translating spoken language into the listener’s preferred dialect, while retaining the speaker’s unique voice, tone, and emotional nuances.

Differing from the existing real-time caption translations in Meet, this feature takes a significant leap by executing direct audio-to-audio conversions. Users will hear the voice of their conversation partner, rather than a mechanical voice, translated into another language with natural vocal characteristics. During a live demonstration, an English speaker and a Spanish speaker easily engaged in conversation, each receiving a translated rendition of the other’s comments in their native language. A subtle version of the original voice plays simultaneously in the background, providing additional clarity and context.

The functionality relies on Google DeepMind’s AudioLM, which has been specifically trained for high-quality speech translation. This results in low-latency dubbing that mimics the presence of a live interpreter during calls, although Google acknowledges there may be minor delays due to processing times.

How to Use Google’s Real-Time Speech Translation

To access the feature, users should navigate to the “Speech translation with Gemini” panel within Meet’s web interface. Here, they can select both the language they are conversing in and the one they wish to comprehend. Once activated, a glowing Gemini overlay and a translation indicator will appear in the upper corner of the screen.

This feature is currently being rolled out in beta for users subscribed to Google’s AI Pro and the newly introduced AI Ultra plans, which is available for $249.99 per month. Notably, for the speech translation feature to function, only one participant in a call is required to have a subscription, enhancing accessibility in group scenarios.

Language Support and Future Developments

Initially, the functionality supports English and Spanish, with expansion to additional languages anticipated in the weeks ahead. Google has also announced that initial trials for business users will commence later this year for Workspace customers.

With Microsoft Teams also looking into AI-driven translations, Google’s initiative highlights a developing trend towards real-time, AI-facilitated communication across different languages.