Google has entered the competitive AI dictation arena with the quiet release of Google AI Edge Eloquent. Currently available on iOS, this offline-first application aims to transform raw speech into polished, professional prose by leveraging advanced on-device processing.
Local Intelligence Meets Professional Editing
At the heart of the app are Google’s Gemma-based automatic speech recognition (ASR) models. Once these models are downloaded, users can dictate without an internet connection, ensuring both privacy and speed. For those seeking even higher quality, an optional “cloud mode” utilizes Google Gemini to further refine and polish the text.
Unlike standard transcription tools that record every verbal stumble, Eloquent is designed to capture intent. It automatically strips out filler words like “um” and “ah” while smoothing over mid-sentence self-corrections. Once the initial transcript is generated, users can instantly reshape it using several transformation presets:
- Key points for quick summaries.
- Formal for professional correspondence.
- Short or Long for varying length requirements.
Tailored for Productivity
The app goes beyond simple transcription by learning the user’s specific vocabulary. It can import jargon, names, and specialized terms directly from Gmail, and users can manually add custom words to its dictionary. For tracking productivity, the app provides a detailed history with search functionality and performance metrics, including words-per-minute and total word counts.
Future Expansion and Integration
While the initial launch focuses on iOS, Google has signaled that an Android version is on the horizon. Documentation suggests the Android release will offer deep system integration, allowing Eloquent to function as a default keyboard. This would enable AI-powered dictation across any text field on the device, accompanied by a floating action button for quick access—a feature set that directly challenges existing tools like Wispr Flow, SuperWhisper, and Willow.
