On August 26, local time, Google launched its latest speech-to-text model, Gemini 3.5 Transcribe. This model not only improves recognition accuracy but also possesses the ability to understand speaker intent, automatically identifying self-corrections, deleting filler words, and converting unstructured spoken language into formatted text. Google calls it its "most accurate" model to date, and it has been integrated into Gboard for Android and the Gemini app for Mac, with a preview also available to developers via the Gemini API. This release coincides with the launch of Gemini 3.5 Live and Gemini 3.5 Live Experimental, together forming Google's new Gemini Audio series. Analysts believe that as voice input increasingly permeates Google products, voice interaction is expected to gradually replace keyboard input, becoming the primary way users connect with AI systems.