On Monday, Google discreetly launched an offline-centric dictation application named “Google AI Edge Eloquent” for iOS, aiming to compete with Wispr Flow, SuperWhisper, Willow, and similar applications.
The application is available for free download, and after the Gemma-based automatic speech recognition (ASR) models are installed, you can begin dictating from your mobile device. Within the application, live transcription is displayed, and when you press pause, the app instinctively eliminates filler words such as “um” and “ah,” refining the text.
Beneath the transcript, there are selections like “Key points,” “Formal,” “Short,” and “Long” to modify the text.
Additionally, you can disable cloud mode for local-only processing. (When cloud mode is activated, the app employs cloud-based Gemini models for text refinement.) Google AI Edge Eloquent can pull in specific keywords, names, and terminologies from your Gmail account if needed. You may also include your custom words in the database.
The app tracks the history of transcription sessions, allowing you to search through all the entries. It can display words dictated in the previous session, measure your words per minute rate, and count the total number of words narrated.
“Google AI Edge Eloquent is a sophisticated dictation application designed to close the gap between natural conversation and professional, ready-to-use text. Unlike typical dictation software that documents hesitations and filler phrases word-for-word, Eloquent leverages AI to grasp your intended message. It automatically removes ‘ums,’ ‘uhs,’ and self-corrections within sentences, yielding clean, precise written content,” states the company’s App Store description.
Currently, the app is exclusively accessible on iOS, but the App Store description alludes to an upcoming Android version. (We have contacted Google for additional details and will provide updates if we receive a response.)
As per the description, Eloquent delivers “seamless Android integration,” allowing it to be set as users’ default keyboard for system-wide availability across any text field. Moreover, the app will feature a floating button function, akin to Wispr Flow’s on Android, for convenient transcription access from any location.
Transcription applications driven by AI are rising in popularity among users as speech-to-text technologies improve. With this experimental app, Google is entering the market. If this trial proves successful, we may witness enhanced transcription capabilities across Android as well.