Google’s latest AI-powered app has introduced Google AI Edge Eloquent, a new dictation tool designed to turn spoken input into clean, ready-to-use text—without relying heavily on cloud processing. What sets the app apart is its ability to refine speech as it’s being transcribed. Instead of capturing every pause or hesitation, it automatically removes filler words such as “um” and “ah,” delivering text that reads more naturally. Once you pause speaking, the app further polishes the output, making it suitable for professional use with minimal editing.
Read More | Iran Signals Potential Strikes on ‘Stargate’ AI Data Centres Amid Escalating Tensions
Built on Google’s Gemma-based automatic speech recognition (ASR) models, the app offers real-time transcription, allowing users to see their words appear instantly on screen. It also includes multiple formatting options—users can quickly reshape their text into “Key points,” or adjust tone and length with modes like “Formal,” “Short,” and “Long.”
For those concerned about privacy or connectivity, the app supports on-device processing. Users can switch off cloud mode entirely, ensuring that transcription happens locally on their device. At the same time, the app can integrate with Gmail to better recognise frequently used names, keywords, and industry-specific terminology. There’s also an option to manually add custom words, improving accuracy over time.
Another practical feature is its session history. The app keeps a record of past transcriptions, including metrics such as total word count and speaking speed. This makes it easier to revisit previous inputs or track productivity patterns over time. At present, Google AI Edge Eloquent is available on iOS, with indications that an Android version is in development. On Android devices, it could eventually function as a system-wide keyboard, enabling dictation across apps and text fields.
Read More | LG Group Eyes Humanoid Robotics Supply Chain with Integrated Component Push
Alongside the app, Google has also introduced Gemma 4, a new open AI model aimed at developers. Unlike proprietary systems such as Gemini or GPT models, Gemma 4 is designed to run locally on a wide range of devices, including smartphones. It combines the flexibility of open models with capabilities typically associated with more advanced, closed systems.
Together, these releases point to a broader push by Google to bring AI capabilities closer to the device—making them faster, more private, and more practical for everyday use.
