Tech giant Google has quietly launched a new offline-first dictation application for iOS, signalling its entry into the rapidly growing market of AI-powered transcription tools. The app, known as “Google AI Edge Eloquent,” is designed to convert speech into polished text, positioning it as a competitor to emerging platforms such as Wispr Flow, SuperWhisper, and Willow.
The application, which is free to download, operates using on-device automatic speech recognition models based on Google’s Gemma framework. Once installed, users can dictate speech and receive live transcriptions, with the system automatically refining the output by removing filler words and correcting sentence structure. Additional features allow users to adjust tone and format through options such as key points, formal style, or shortened summaries.
Further enhancing its functionality, the app supports both offline and cloud-based processing. When cloud mode is enabled, it integrates with Gemini to improve text quality, while local-only mode ensures privacy by keeping processing on the device. It can also personalise transcription by importing frequently used terms from Gmail accounts and allowing users to add custom vocabulary, alongside tools for tracking word count, speaking speed, and transcription history.
Although currently limited to iOS devices, the app’s listing hints at a broader rollout, including future integration with Android systems where it could function as a default keyboard for system-wide dictation. The move reflects Google’s growing interest in AI-driven productivity tools, as demand for more accurate and seamless speech-to-text solutions continues to rise globally.