New Delhi: Google has recently started rolling out its enhanced Gemini AI model to bolster the capabilities of its translation services for users. The company has stated that it is integrating its state-of-the-art Gemini AI model directly into Google Translate, claiming to offer significantly smarter, more natural text translations. The company is also bringing a latest real-time speech-to-speech beta, which is especially designed for use with the headphones. This newest rollout began recently in the US and India, which supports English and nearly 20 other languages, including Spanish, Hindi, Chinese, Japanese and German.
Google Translate is now leveraging Gemini’s advanced reasoning capabilities to move beyond literal, word-for-word translations. By parsing context, the AI can now accurately translate complex idioms like stealing my thunder into natural-sounding phrases in other languages, such as Spanish. Google also states it is featuring a beta experience for live translation through headphones that builds on Gemini’s native speech-to-speech capabilities. According to the company, the model is designed to preserve the speaker’s original tone, emphasis and cadence, making it easier to follow the conversations or lectures.
Additionally, Google is expanding its language learning features within the Translate application to nearly 20 new countries, including Taiwan, Sweden and Germany. The application now offers improved feedback based on the user’s speaking practice, providing helpful tips for refinement. Users can now track streaks, which are the number of consecutive days they have practised, to help maintain consistency and reach learning goals. Latest practice sets include English to German and Portuguese, as well as multiple languages to English. Google intends to bring these live translation features to iOS and more countries in 2026.
