Microsoft is narrowing the feature gap between Google’s flagship hardware and the broader Android ecosystem. A new “AI voice” feature currently in testing within the Microsoft SwiftKey beta brings the core functionality of the Pixel 11’s “Rambler” mode to third-party devices, allowing users to dictate messy, unfiltered speech that the keyboard then converts into clean, structured text.
The feature, discovered in SwiftKey beta version 9.13.16.4, is designed to eliminate the hesitation common in standard voice-to-text. By utilizing an on-device AI model, the keyboard identifies and removes filler words like “um” and “uh,” corrects mid-sentence stammers, and reformats rambling sentences into coherent prose. This functionality directly mimics the Rambler feature that served as a headline software addition for Google’s latest Pixel hardware.
Offline Processing and Technical Requirements
Unlike some cloud-reliant AI tools, SwiftKey’s implementation prioritizes privacy and speed through local processing. Upon enabling the feature, the app requires a one-time download of a 163MB AI model. Once installed, the SwiftKey AI voice mode operates entirely offline, ensuring that voice data does not leave the device for transcription or refinement.
This on-device approach allows the feature to function on a variety of high-end hardware beyond Google’s ecosystem. Early testing has confirmed the feature is fully operational on contemporary flagship hardware, such as the Samsung Galaxy Z Fold 8. This democratization of AI-assisted dictation provides a viable alternative for users who prefer non-Google hardware but desire the “smart” transcription capabilities previously locked to the Pixel 11 series.
Functional Differences from Pixel’s Rambler
While SwiftKey successfully replicates the “cleanup” aspect of Rambler, there are distinct technical trade-offs when compared to Google’s native Gboard integration. According to reported tests of the beta, SwiftKey does not provide live transcription; instead of seeing words appear as they are spoken, users must finish their thought before the keyboard processes the audio and displays the final, polished text.
Furthermore, SwiftKey’s current iteration lacks the deep system-level commands found in the Pixel 11 version. Users cannot currently insert emojis via voice commands or instruct the AI to adjust the tone or style of the text immediately after dictation. In contrast, Google’s Rambler leverages tighter integration with Gemini to allow for post-speech editing and stylistic shifts.
Despite these omissions, the new SwiftKey feature represents a significant shift in how AI-driven accessibility tools are distributed. By packaging the technology into a downloadable keyboard, Microsoft is effectively decoupling one of the Pixel’s most marketed software advantages from its proprietary hardware.
