On April 6 (Monday) local time, Google quietly launched an experimental voice input app called "Google AI Edge Eloquent" on the iOS platform. The app emphasizes "offline-first" and "intelligent polishing," aiming to use edge-side AI technology to convert natural spoken language into professional and concise text in real time. This move marks Google's official entry into the high-end AI speech-to-text market led by Wispr Flow and SuperWhisper.
Core Technology and Features:
Eloquent is equipped with Google's latest released Gemma4 series (E2B/E4B specifications) automatic speech recognition (ASR) model. This model supports fully offline operation, allowing users to perform local transcription after downloading the model package, effectively protecting privacy and reducing latency. The app features a powerful "intelligent noise reduction" function that automatically identifies and filters filler words like "um" and "uh" as well as repeated corrections, producing logically coherent text.
Product Integration and Interaction:
Multimodal Style Conversion: Offers four text processing modes: "Highlights," "Formal," "Concise," and "Complete."
Cloud Collaboration (Optional): When cloud mode is enabled, the app will use the cloud-based Gemini model for deep text cleaning.
Personalized Context: Supports importing users' unique keywords, names, and terminology from Gmail, and allows building a custom vocabulary list.
Productivity Statistics: Displays real-time dictation word count, words per minute (WPM), and historical session records.
Market Strategy and Future Plans:
Currently, the app is available for free
