Tech giant Google has quietly introduced a new offline first dictation application, “Google AI Edge Eloquent,” on iOS, signalling its growing push into the competitive speech to text market. The free app is designed to rival emerging players such as Wispr Flow, SuperWhisper and Willow, offering users a more refined and intelligent transcription experience powered by on device artificial intelligence.
Built on Gemma based automatic speech recognition models, the app allows users to dictate speech and instantly view live transcriptions, even without internet connectivity. Once paused, the system automatically removes filler words and restructures sentences into polished text, reflecting a shift from raw transcription to meaning driven writing. Additional tools enable users to refine outputs into formats such as key points, formal tone, or varying lengths, making the app useful for both casual and professional communication.
A key feature of the application is its hybrid processing capability, allowing users to toggle between local only processing and cloud enhanced performance. When connected, the app leverages advanced Gemini models to further clean up text, while still giving users the option of full offline privacy. It also integrates with services like Gmail to import commonly used words, names and jargon, alongside custom vocabulary settings tailored to individual users.
Beyond transcription, the app provides productivity insights such as speech speed, word count and searchable history of past sessions, reflecting a broader trend toward intelligent personal productivity tools. Though currently limited to iOS, indications of an upcoming Android version suggest deeper integration across devices. Industry observers say the move positions Google to strengthen its foothold in AI powered communication tools, as demand for faster and more accurate voice driven interfaces continues to rise.