Google has introduced a new dictation app called Google AI Edge Eloquent on iOS. The app focuses on offline speech-to-text, putting it in competition with tools like Wispr Flow, SuperWhisper, and Willow. It is free to download, and once users install the required Gemma-based speech recognition models, they can start dictating without needing an internet connection.
The app provides live transcription as you speak. When you pause, it cleans up the text by removing filler words like “um” and “ah.” It also refines the output into smoother, more readable sentences. Users can further adjust the text using built-in options such as Key Points, Formal, Short, and Long, depending on the desired tone or format.
One of its key features is flexibility between offline and cloud processing. Users can switch off cloud mode to keep everything on-device. If cloud mode is enabled, the app uses Google’s Gemini models for deeper text refinement. This gives users a balance between privacy and advanced processing.
The app can also personalize results. It can pull in names, keywords, and jargon from Gmail, or users can manually add custom terms. Over time, this helps improve accuracy, especially for professional or niche vocabulary.
Eloquent also keeps track of past sessions. Users can view their transcription history, search through previous entries, and check useful stats like words per minute and total word count. This makes it more than a basic dictation tool—it becomes a productivity tracker too.
Although the app is currently limited to iOS, references in its listing suggest an Android version is in development. Future support may include system-wide access, such as setting it as a default keyboard or using a floating button for quick dictation.
With speech recognition improving fast, tools like Eloquent reflect the growing demand for accurate and private voice typing. If this early release performs well, similar features could expand across Google’s broader ecosystem.
