Google has launched AI Edge Eloquent, a new application focused on voice-to-text transcription. The application emphasizes privacy, processing voice data directly on the device, which means your spoken words are not stored by Google when cloud mode is disabled. This offline-first approach positions the free tool as a direct competitor to existing paid services like Willow and Wispr Flow.
On-Device Processing and Enhanced Accuracy
The core functionality of AI Edge Eloquent revolves around its ability to perform speech-to-text conversions with a strong emphasis on privacy. By downloading speech recognition models, like those based on Gemma, directly to the user's device, the app aims to provide a secure dictation experience. This local processing model offers an alternative for users concerned about their voice data being uploaded to cloud servers. The app also features automatic text refinement and cleanup, a feature designed to produce transcriptions that closely resemble polished written text rather than raw dictation.
Read More: Simple API Testing Tools Needed for Quick Checks, Not Complex Ones
Tailoring the Transcription Experience
AI Edge Eloquent offers several features to customize and improve the transcription output. Users can enhance accuracy by:
Importing Keywords: Connecting the app to their Gmail account allows it to import names, jargon, and specific terminology.
Custom Dictionaries: Users can manually add their own frequently used words or phrases to a dedicated dictionary.
Furthermore, the app includes built-in transformation tools that can adjust the generated text's tone and structure. These tools can format the text into bullet points, or modify it to be more formal, concise, or expansive. A live waveform visualizes the user's speech, and a real-time transcription appears on screen as they speak. Past dictations are accessible via a History tab.
Platform Availability and Future Prospects
Currently, Google AI Edge Eloquent is available on iOS. However, references within the App Store description indicate that an Android version is planned. The application operates similarly to Google's existing 'AI Edge Gallery' app, which allows for local execution of generative AI models.
This release signals Google's increasing foray into providing AI-powered tools that operate locally, a trend mirroring efforts to enhance user privacy and reduce reliance on constant cloud connectivity. The app's free access and robust offline capabilities make it a notable entry in the speech-to-text market.
Read More: Docker Model Runner lets users run AI models on their own computers easily
