Google has quietly stepped into the fast-growing AI dictation space with a new offline-first app for iPhone users called Google AI Edge Eloquent. Designed to compete with popular speech-to-text tools like Wispr Flow, SuperWhisper, and Willow, the app focuses on turning everyday speech into clean, usable writing—without requiring a constant internet connection.
Google AI Edge Eloquent is free to download on iOS. Once you install it and download its on-device automatic speech recognition models (built on Google’s Gemma technology), you can start dictating directly on your phone. As you speak, the app shows a live transcription in real time. Then, when you pause, it goes a step further than standard dictation apps: it automatically removes filler words such as “um” and “ah,” and it polishes the output so the text reads more naturally.
After your transcript appears, Eloquent also gives you built-in rewriting options to quickly reshape what you said. You can choose tools like “Key points,” “Formal,” “Short,” and “Long” to transform the text depending on whether you need a quick summary, a more professional tone, or a more expanded version of your thoughts. This makes it useful for everything from meeting notes and emails to journaling and drafting content on the go.
A major selling point is flexibility around privacy and processing. Users can disable cloud mode and keep everything local-only, which means your transcription can stay on-device. When cloud mode is enabled, the app uses cloud-based Gemini models to help with text cleanup, offering an upgraded processing option for those who want it.
Eloquent also aims to improve accuracy with smarter vocabulary support. If you choose, it can import certain keywords, names, and jargon from your Gmail account, helping the app better recognize the terms you use most often. You can also add custom words manually—handy for specialized work vocabulary, unusual names, or industry acronyms that typical dictation apps often get wrong.
On top of transcription and rewriting, the app includes session tracking features that make it feel more like a productivity tool than a basic speech-to-text recorder. It keeps a searchable history of your dictation sessions and can display stats such as the words dictated in your last session, your speaking speed in words per minute, and your total word count.
Google’s App Store description frames the app as a bridge between natural speech and professional writing, emphasizing that it’s meant to capture your intended meaning—not every hesitation, filler phrase, or mid-sentence correction that usually sneaks into raw transcriptions.
For now, Google AI Edge Eloquent is only available on iOS, but the listing suggests an Android version is being considered. The description also hints at features aimed at deeper Android integration, including the ability to set it as a default keyboard for dictation across any text field, plus a floating button for quick access to transcription from anywhere—similar to the shortcut approach seen in other Android dictation apps.
AI transcription apps are surging in popularity as speech recognition models improve, and Google’s experimental launch shows it’s taking the category seriously. If the iOS test lands well, it could be an early sign of more advanced dictation and transcription features coming to Android in the future, potentially making speech-to-text a more powerful everyday writing tool across mobile devices.
