Google Introduces Gemini 3.5 Transcribe with Smart Jargon Recognition

Google has expanded its generative audio portfolio with the launch of Gemini 3.5 Transcribe, a specialized speech-to-text model designed to convert spoken language into clean, formatted text in real time. The model automatically filters out filler words, pauses, and disfluencies like 'ums' and 'ahs' while preserving conversational nuance.

Engineered to support over 85 global languages, Gemini 3.5 Transcribe features specialized contextual modules capable of recognizing complex medical, legal, and software engineering jargon. The underlying technology powers Gboard's new 'Rambler' voice input feature and is being integrated across Google Chrome and Workspace apps.

By combining acoustic signal processing with deep transformer reasoning, Google aims to set a new benchmark for automated transcription accuracy in noise-heavy operational environments.