Google deployed Gemini Live in France and in French this Thursday, October 3. The feature is no longer reserved for Pixel smartphones, but is available on all Android models. This new kind of voice assistant will soon be deployed on iOS.
Gemini Live has been available in English-speaking countries since the release of Pixel 9, Google’s latest smartphones. This new kind of voice assistant, capable of speaking fluently, speaks French perfectly… despite its unavailability in France. You have to switch your smartphone to English to activate it, then speak to it in French.
Thursday October 3, Google announced the end of this differentiation: Gemini Live is now available in France, with native French support. This is the rival of the new ChatGPT Voice Advanced, currently blocked in the European Union.
A much stronger voice assistant than Google Assistant
Since the announcement of Assistant with Bard almost a year ago (two months before the Gemini announcement), the planned death of Google Assistant has seemed inevitable. Google is gradually merging these two voice assistants, to allow Gemini, which is based on an LLM (large language model), to perform classic actions (such as a timer or a call).
With Gemini Live, the company goes even further. We can discuss in a way “ natural » with the Google chatbot, for free.
BFM Tech&Co was able to test the functionality, which allows you to choose from ten voices (five female and five male). The Gemini application sees an icon appear at the bottom right of the screen to start the interaction with the chatbot, as if it were a voice call. We can therefore discuss with him, ask him questions, interrupt him, ask him for details, as if we were writing to Gemini. This is the same interface as in the English version which, remember, already spoke French.
Françoise Beaufays, researcher at Google in charge of machine learning for voice models, interviewed by our colleagues, specifies that “ The basic model is multilingual and optimized for the languages presented to users. Everything is adapted for French: voice recognition, the model, the responses it produces and obviously the voices “. It is even possible to change languages on the fly. Eventually, it should be possible to ask Gemini Live to analyze documents and photos, as is the case for the currently written version. At the end of a discussion, Gemini writes a summary of the discussion you had with him.
Gemini Live is being deployed on Android smartphones, but this is an operation that will take several weeks, as Google indicated. If you don’t have access to the feature yet, that’s completely normal: just wait.
The voice chatbot war is on
At the same time, OpenAI launched ChatGPT’s new voice mode, Voice Advanced, based on GPT-4o, a few days ago. It is still not available in France, but can be tested via a VPN by virtually locating in the United States. He also offers great speaking skills, with emotion in his voice and the ability to understand your intonations.
On October 1, Microsoft also announced a whole bunch of new features for its chatbot, Copilot, based on GPT. Among them, Copilot Voice, which aims to facilitate “ interacting with your AI companion “. Microsoft is using the same technology as OpenAI to reinvent voice conversations, with the bet that you will soon be “phoning” its voice assistant for hours.
OpenAI, Google and Microsoft are positioning themselves on voice interactions with chatbots based on LLMs, while Apple is refining its new Siri (also with an LLM, but with dedicated tasks). The future of AI could involve the advent of these new voice assistants.
Related News :