An Android Input Method Editor (IME) providing offline voice recognition and translation using the Whisper engine.
WhisperIME is an Android Input Method Editor that uses the Whisper speech‑recognition engine to convert spoken language into text. It solves the need for private, offline voice typing and translation on Android devices, without relying on cloud services. The app downloads models once and then works completely locally, ensuring data never leaves the user's device.
Android users who require offline voice‑to‑text input, value privacy, and want an alternative to built‑in voice recognition services. It's also suitable for developers looking to integrate offline speech recognition into their apps.
Developers choose WhisperIME for its strong privacy guarantees, full offline operation, and flexibility as both an IME and a system‑wide voice input service. It provides a transparent, open‑source alternative to proprietary voice recognition tools.
Android Input Method Editor (IME) based on Whisper
Open-Awesome is built by the community, for the community. Submit a project, suggest an awesome list, or help improve the catalog on GitHub.
After the initial model download (~435 MB), all speech processing occurs locally with no internet connection required, ensuring user data never leaves the device.
Functions as a system-wide IME, standalone translation app, and RecognitionService, supporting calls via RecognizerIntent for broad app compatibility.
Offers a fast English-only model for speed and a comprehensive multi-lingual model for broader language support, allowing users to balance performance versus functionality.
Built on MIT-licensed components like Whisper-Android and OpenAI Whisper, providing clear attribution and community-driven development.
The README explicitly states it will stop working on certified Android devices after 2026/2027 due to Google's policy changes, limiting long-term viability.
Activating as system-wide voice input requires accessing separate Android settings and potentially using ADB shell commands if not listed, adding complexity.
The multi-lingual model is noted as 'much slower,' and there's a 30-second recording limit per session, hindering longer or real-time speech tasks.
Whisper is an open-source alternative to the following products:
Google Voice Typing is a speech recognition feature that allows users to dictate text using their voice in Google Docs and other Google Workspace applications.
Samsung Voice Input is a speech recognition feature on Samsung devices that allows users to input text and control their device using voice commands.