Google has just quietly launched a new spell-reading application called Google AI Edge Eloquent, marking a remarkable step forward in the technology race to convert voice into text.
The biggest difference of this application is its offline capability, helping users without internet connection can still use core features.
The application is currently released for free on iOS. After downloading and installing Gemma-based voice recognition (ASR) models, users can start reading the spelling directly on their phone.
During use, Eloquent displays the record in real time. When the user pauses, the system will automatically process the content, remove buffer words like "uhm", "a" and edit the sentence to make it more coherent.
This is a difference compared to many traditional tools that only record verbatim speech.
The application also provides options such as "Main Ideas", "Significant", "Short" or "Long" for users to quickly modify text according to their needs, from quick notes to professional content.
One of the notable points is that Eloquent allows users to completely turn off cloud mode. At that time, all data will be processed directly on the device, increasing privacy and security.
If online mode is turned on, the application will take advantage of Gemini models on the cloud platform to optimize text better.
Users can allow the application to learn keywords, aliases or specialized terms from their Gmail accounts, or add custom words to improve accuracy.
The application also saves the history of spelling reading sessions, allowing re-searching of content. Statistics such as word count, read speed (word/minute) or total number of spoken words are also displayed, providing a detailed tracking experience.
The appearance of Eloquent shows that Google is directly competing with AI dictation applications such as Wispr Flow, SuperWhisper or Willow.
In the context of increasingly accurate voice recognition models, this market is developing rapidly.
Although currently only available on iOS, the application description shows that Google has prepared for the Android version.
Even, this version is expected to be deeply integrated into the operating system, allowing use as a default keyboard or through floating buttons for quick access to all applications.
Google's focus on offline processing reflects a new trend when bringing AI from the cloud directly to user devices.
This not only helps speed up processing but also reduces risks related to personal data.
If the test is successful, smart spell reading features may soon become the standard on Android, changing the way users interact with mobile devices, from typing to natural speech.
