The live translation revolution: Google is expanding its use of artificial intelligence and integrating the Gemini model into the Google Translate app, with a new feature that enables real-time translation directly to headphones. This capability brings users closer to an almost instantaneous cross-language communication experience without the need to read or type.
The new feature, called Live Translate, allows users to hear a live translation of a conversation in a foreign language directly through their headphones. After opening the app, selecting the languages, and connecting the headphones, users can activate translation mode and listen to real-time translations of the person speaking to them. The technology also works when watching content such as movies or series in a foreign language, allowing users to understand dialogues while viewing.
Google explains that the main innovation lies not only in translating words, but in Gemini’s ability to preserve the speakers’ intonation, rhythm, and emphasis. According to the company, this creates a more natural translation experience that allows users to follow conversations in a way that is closer to the original language.
The system supports more than 70 languages, including English, Arabic, French, German, Spanish, Hindi, and Chinese. The feature can also be used without headphones, but the core experience is designed for direct listening, reducing the need to look at the screen and enabling smoother communication.
Until now, the service was mainly available to Android users in a limited number of countries, but Google is now expanding the rollout to iPhone devices and additional countries across Europe, Asia, and Africa. Among the countries where the feature is already active are Germany, France, Spain, Italy, the United Kingdom, and Japan.
Google’s move places it in direct competition with Apple, which has already introduced real-time translation capabilities in AirPods on advanced models. However, Google’s advantage lies in the advanced integration of artificial intelligence models like Gemini, which enable not only translation but also more sophisticated natural language processing.
Despite the strong promise, the use of the feature still depends on technological conditions such as internet connection quality, background noise, and differences between speakers. Translation accuracy may also vary depending on the language, accent, and complexity of the conversation.