Google launches Gemini 3.5 Translate for real‑time speech

Google has launched a new artificial intelligence tool designed to enable near-instantaneous, natural-sounding conversations between speakers of different languages. The company announced Gemini 3.5 Live Translate on June 9, 2026, describing it as its most advanced speech-to-speech translation model to date.

The core innovation of the model is its ability to process and translate speech continuously, without waiting for a speaker to finish a sentence. This continuous stream translation results in a delay of only a few seconds, which Google compares to the lag on a long-distance phone call. The system automatically detects which of more than 70 supported languages a person is speaking, enabling thousands of possible language pairings without manual setup.

According to Google, the model is designed for real-world use, capable of filtering out background noise and handling overlapping voices and informal speech. It also aims to preserve a speaker's authenticity by matching their pacing, intonation, and emotional tone in the translated audio, moving beyond robotic, synthetic voices. Company officials stated the technology is intended for practical applications such as customer...