Google has quietly introduced a new AI-driven dictation application called Google AI Edge Eloquent. Designed to compete with specialized tools like Wispr Flow and SuperWhisper, the app aims to transform spoken word into polished, professional text by leveraging advanced on-device artificial intelligence.
How It Works: From Speech to Polished Prose
Unlike traditional dictation software that simply transcribes every sound—including stumbles and hesitations—Eloquent uses Gemma-based automatic speech recognition (ASR) models to interpret intent.
The app’s core functionality includes:
– Automatic Refinement: It intelligently filters out filler words such as “um” and “ah,” as well as mid-sentence self-corrections.
– Contextual Transformation: Once a transcription is complete, users can instantly reformat the text using presets like Formal, Short, Long, or Key Points.
– Personalized Vocabulary: The app can import specific names, jargon, and keywords from a user’s Gmail account or allow for manual entry of custom words to improve accuracy.
Privacy and the “Offline-First” Approach
A standout feature of Eloquent is its emphasis on local processing. While the app offers a “Cloud Mode” that utilizes Google’s powerful Gemini models for enhanced text cleanup, users can opt to turn this off entirely.
By downloading the necessary AI models directly to the device, the app can function offline. This “offline-first” architecture is a significant trend in the AI industry, as it addresses growing user concerns regarding data privacy and the security of sensitive information being sent to remote servers.
Features and User Interface
The app provides a comprehensive dashboard for managing transcribed content, including:
– Live Transcription: Users can watch the text appear in real-time.
– Session History: A searchable archive of all previous transcriptions.
– Performance Metrics: Data on words-per-minute (WPM) and total word counts to help users track their dictation efficiency.
Availability and Platform Confusion
Currently, the app is available for download on iOS, though the rollout appears somewhat inconsistent. While the App Store listing mentions an Android version, the app is not yet live on that platform.
However, the description suggests significant features are in development for Android users, including:
– System-wide Integration: The ability to set Eloquent as the default keyboard.
– Floating Access: A floating button (similar to Wispr Flow) that allows users to trigger transcription from any text field across the operating system.
“Eloquent utilizes AI to capture your intended meaning… outputting clean, accurate prose,” according to Google’s official description.
Conclusion
Google AI Edge Eloquent represents a shift toward more intelligent, private, and “clean” transcription tools. By combining on-device AI with seamless text refinement, Google is positioning itself to compete in the growing niche of high-productivity, privacy-conscious AI utilities.
