GoKawiilGoogle adds Gemini 3.5 Transcribe with filler-word removal and voice editing
Google has released three new Gemini Audio models: 3.5 Transcribe, 3.5 Live, and 3.5 Live Experimental. Transcribe automatically strips out filler words like 'um' and 'uh', formats text, adapts to custom vocabulary and jargon, and can identify up to three speakers with word-level timestamps. The Live models improve handling of interruptions, multilingual recognition, and real-time reasoning narration for voice chat.