Google Unveils Offline AI App That Converts Speech into Professional Text

Google has unveiled a new experimental dictation application designed to convert spoken words into polished, professional text, even without an internet connection.

The app, named Google AI Edge Eloquent, has been quietly released on iOS as part of the company’s broader push into on-device artificial intelligence. Built on Gemma-based automatic speech recognition (ASR) models, the tool allows users to dictate content offline once the models are downloaded, offering a seamless speech-to-text experience without relying on cloud connectivity.

Unlike conventional dictation tools that transcribe speech verbatim, Eloquent focuses on delivering refined output. It automatically removes filler words such as “um” and “ah,” corrects interruptions, and restructures sentences to produce clear, professional-grade text aligned with the speaker’s intent.

Users can view live transcriptions as they speak and later refine the content using built-in options such as “Key points,” “Formal,” “Short,” and “Long.” These features enable quick transformation of raw speech into summaries, structured notes, or formal written content, making the app particularly useful for professionals and content creators.

While the app prioritizes offline functionality, it also offers an optional cloud mode powered by Google Gemini. When enabled, this mode enhances text processing and cleanup using more advanced AI models. However, users can choose to keep all processing local, reinforcing privacy and control over personal data.

Eloquent also supports personalization features, allowing users to add custom words, names, and technical jargon. It can optionally import relevant vocabulary from Gmail to improve transcription accuracy. Additionally, the app tracks metrics such as words-per-minute speed, total word count, and maintains a searchable history of past dictation sessions.

The launch places Google in competition with a growing number of AI transcription tools, including Wispr Flow, SuperWhisper, and Willow, as advancements in speech recognition continue to reshape digital productivity.

Although currently limited to iOS, references within the app’s description suggest that an Android version is in development. The upcoming release is expected to offer deeper system integration, including the ability to function as a default keyboard and provide system-wide dictation access.

Also read:

Google Chrome’s New Update Will Make Videos and Audio Load Faster on Websites