Google officially unveiled a groundbreaking real-time speech translation feature for Google Meet during the Google I/O 2025 conference. Powered by advanced large language audio models developed by Google DeepMind, this technology enables natural, fluid conversations between participants speaking different languages without the typical barriers of traditional translation tools.
Preserving Nuance and Tone
Unlike standard text-based translation, this new Meet capability focuses on preserving the speaker’s original voice, tone, and emotional expression. When a participant speaks, the listener hears the original audio faintly in the background, with the translated speech layered seamlessly on top. This maintains the human element of communication, ensuring the speaker’s intent is not lost in digital conversion.
Transforming Global Communication
Google highlights a wide array of practical applications for this technology. Beyond personal use—such as enabling seamless communication between multilingual family members—the feature is designed to bridge gaps in global business operations. By providing low-latency translation, the platform now supports multi-party conversations that were previously impossible, allowing diverse teams to collaborate effectively in real time regardless of their native language.
Rollout and Language Support
The feature enters beta testing this Tuesday for consumer AI subscribers. Initial language support will focus on English and Spanish, with Google promising to expand to Italian, German, and Portuguese in the coming weeks.
Enterprise Expansion
While the initial rollout targets individual users, Google is actively developing the infrastructure for enterprise environments. Early testing of speech translation for Workspace customers is scheduled to commence later this year, marking a significant step toward universal accessibility in professional digital meetings.