Talking to your phone just got a lot smarter and a lot quieter on the data front. Google has quietly released a new AI dictation app that works even when you are offline, turning messy speech into polished text without sending your voice to the cloud.
The app, called Google AI Edge Eloquent, reflects a growing shift in how AI tools are built and deployed. Instead of relying entirely on remote servers, Google is pushing intelligence onto the device itself, signaling a new phase for on-device AI.
Offline-first design emphasizes privacy
TechCrunch reported that Google designed the app to work offline using Gemma-based speech recognition models, allowing users to dictate text directly on their devices. Once models are downloaded, no internet connection is required for core functionality.
“Google AI Edge Eloquent is an advanced dictation app engineered to bridge the gap between natural speech and professional, ready-to-use text,” the company said in its App Store description. “It automatically edits out ‘ums,’ ‘uhs,’ and mid-sentence self-corrections.”
9to5Google also noted that the app includes a “fully offline” mode where conversations never leave the device, reinforcing its privacy-focused design.
Users can optionally enable cloud-based Gemini models for additional text polishing, creating a hybrid approach between local and cloud AI.
Features go beyond basic transcription
Google AI Edge Eloquent aims to do more than convert speech to text. TechCrunch highlighted that it restructures spoken input into readable formats and offers editing tools such as “Formal,” “Short,” “Long,” and “Key points.”
The app also includes:
- Real-time transcription with automatic cleanup
- Clipboard-ready output for quick sharing
- Usage metrics like word count and speaking speed
Users can customize vocabulary by importing jargon from Gmail or adding their own terms, improving accuracy for specialized workflows. The app also stores transcription history with search and deletion options.
According to 9to5Google, Google positions the tool as “voice dictation without subscriptions” with no usage caps, differentiating it from competitors that are experimenting with paid tiers.
Availability and limitations
The app is currently available on iOS and supports English only. Google’s developer documentation said that availability in regions such as the UK, Switzerland, and the EEA is limited due to regulatory requirements, with expansion planned.
Google also said it is evaluating additional platforms, including desktop environments, for future releases. Early indications suggest deeper Android integration could follow, potentially enabling system-wide dictation access.
For now, the release appears experimental. But it signals Google’s broader ambition to embed AI directly into everyday workflows while reducing dependence on constant connectivity.
Discover six Google Gemini features most people haven’t discovered yet and how they can boost productivity.