Google’s new AI app turns speech into text in real time—even offline: Know what it is

Google has introduced a revolutionary speech-to-text application named Google AI Edge Eloquent, designed to function seamlessly offline on iOS devices. Utilizing the advanced Gemma AI model, this innovative app promises to deliver precise speech recognition, transcribing spoken words into text in real time. As users communicate, they will benefit from live transcriptions, complete with features that eliminate pauses and fillers, ensuring a clean and accurate output. So, what exactly is Google AI Edge Eloquent? It stands out as a specialized app intended for immediate speech-to-text conversion. According to Google’s app store description, Eloquent serves as an "advanced dictation app" that reconciles natural speech with polished, professional-grade text. By leveraging AI technology, the app refines spoken language, removing unnecessary words and errors to produce high-quality transcripts. How does this application operate? After downloading Google AI Edge Eloquent, users must install the Gemma-based automatic speech recognition (ASR) models to get started. A major feature is the option to disable cloud mode, allowing transcriptions to occur directly on the device, ensuring privacy and offline functionality. When in cloud mode, the app employs Gemini to enhance and streamline the text further. Additionally, it learns user-specific vocabulary by importing terms and jargon from Gmail, allowing for a personalized experience, while users can also input custom words. The app not only records transcriptions but also maintains a history, enabling users to track metrics such as transcription speed and total word count. With its entry into the market, Google AI Edge Eloquent positions itself as a competitor to existing transcription applications like Wispr Flow, SuperWhisper, and Willow. Currently, Google AI Edge Eloquent is available for iOS devices, with an anticipated Android version on the horizon. The app is designed for "seamless integration," functioning effortlessly across various systems without the need for app switching. Moreover, it features a convenient floating button akin to Wispr Flow, allowing users to initiate transcription swiftly from any screen location.