Google Launches AI Edge Eloquent Dictation App to Advance On-Device Speech Recognition

Google has introduced Google AI Edge Eloquent, a new AI-powered dictation app on iOS designed to improve speech-to-text accuracy using on-device processing. The app leverages Gemma-based automatic speech recognition models to deliver offline transcription, while optional cloud integration enables enhanced text refinement through Gemini models.

The application provides real-time transcription and automatically removes filler words, restructuring speech into cleaner, more professional text. It also offers transformation features such as summarisation, tone adjustment, and length variation, alongside session history tracking and performance insights including speaking speed and word count.

Google positioned the app as part of a broader shift toward AI-native interfaces, enabling natural language input across devices. While currently limited to iOS, the company has indicated that expanded functionality, including deeper integration and broader platform availability, is in development.