Google announces Gemini 3.5 Transcribe for AI-powered speech-to-text

Google has introduced Gemini 3.5 Transcribe, an AI speech-to-text model that improves speed and accuracy over its predecessor Chirp 3. It removes filler words and corrects speech in real-time, supporting 85 languages and multiple speakers. The model is currently used in Gboard's Rambler feature on Pixel 11 and will expand across Google's services.