[Google Launches Gemini 3.5 Transcribe Speech-to-Text Model, Capable of Automatically Removing Filler Words] According to IT Home, Google has launched the Gemini 3.5 Transcribe speech-to-text model, designed to optimize the voice input experience. It can filter out filler words in real-time, automatically correct verbal slips, and supports 85 languages as well as audio conversations involving up to three participants. The feature is currently enabled in the 'Rambler' function of the Gboard keyboard on Pixel 11 devices and will be expanded to more services in the future. Compared to its predecessor Chirp 3, its transcription speed has improved by approximately 70%, and the real-time error rate has dropped to 5.5%. The model also supports custom vocabularies, but due to its semantic understanding and text refinement capabilities, it may alter the original phrasing, making it unsuitable for scenarios requiring strict preservation of original wording.