Google has announced Gemini 3.5 Transcribe, a new speech-to-text model designed to refine voice input by removing filler words and corrections. The model already powers the Gboard Rambler feature on the Pixel 11 and will soon roll out across Google's ecosystem.
Google says it transcribes speech about 70 percent faster than its previous engine, Chirp 3, and has cut the live-speech error rate to 5.5 percent from 7.32 percent. The improvements aim to reduce the hassle of fixing typos when using voice dictation.
Comments