Google announces Gemini 3.5 Live Translate for instant voice-to-voice translation

Google has long been at the forefront of real-time translation technology, marking it as one of its key machine learning initiatives. Over the years, the tech giant has showcased various demonstrations at its events, but previously, access was limited to specific devices like Google phones and earbuds. Now, with the launch of Gemini 3.5 Live Translate, Google is significantly broadening the scope of its translation services. This latest iteration promises instant voice translations with reduced latency, making it more accessible than ever. Gemini 3.5 is part of the version 3.5 family that debuted during the I/O event. Previously, only the Flash version was available, but a Pro model is anticipated to be released shortly. The Gemini 3.5 Live Translate feature is a sophisticated speech-to-speech model capable of automatically detecting and translating over 70 languages. Google claims that this new technology is proficient enough to keep pace with everyday conversations, translating just moments after the speaker while also mimicking their intonation, pacing, and pitch. This results in a more natural-sounding voice, distinguishing itself from typical robotic outputs. The initial demonstrations, conducted under controlled settings, show promising results. Soon, users will be able to experience the capabilities of Gemini 3.5 Live Translate firsthand, as it rolls out across various Google platforms. Additionally, developers can start working with a public preview available through the Gemini Live API or AI Studio. This model continuously processes speech and automatically manages multilingual inputs, relieving developers from the need to adjust settings manually. Furthermore, it effectively filters out background noise, enhancing performance in noisy environments.