On April 6, 2026, Google rolled out an experimental voice input application, Google AI Edge Eloquent, on the iOS platform. This app is designed with a strong emphasis on offline capabilities and is underpinned by Google's lightweight model, Gemma. Once users download the corresponding ASR (Automatic Speech Recognition) model, they can enjoy seamless offline operation without any interruptions. Its key features encompass:

• Real-time transcription with automatic elimination of filler words such as 'um' and 'ah', ensuring a clean and concise transcript.
• Real-time text refinement during the transcription process, transforming spoken language into coherent and logical written form.
• Support for importing custom vocabulary, thereby enhancing the recognition accuracy of rare or specialized terms.
• Multi-format export options post-transcription, coupled with an integrated conversation history management system that facilitates swift retrieval of past recordings through keyword searches.

Users have the flexibility to opt for local computing power for data processing, thereby safeguarding their privacy and security. Alternatively, they can leverage the cloud-based Gemini model to refine and optimize the text quality further. At present, the application is exclusively available on the iOS platform. However, Google officials have confirmed that Android compatibility is on the horizon, promising a broader user base in the near future.