Official documentation
woheller69/whisperIME: Android Input Method Editor (IME) based on Whisper
woheller69, GitHub, n.d.
Read the originalRead on October 19, 2026
- What it is
- The project readme for Whisper IME, an Android keyboard and voice input app built on OpenAI's Whisper models. It explains the one-time model download, says recognition runs offline after that, and describes use as an input method editor or as system-wide voice input.
- How much weight it carries
- Written by the project maintainer and published in the public repository, so it shows what the project says it does. It is a volunteer project rather than a company, and the readme gives no benchmark, version or device list for its claims.
Facts taken from this source
- Whisper IME's readme says voice recognition works entirely offline once the models are downloaded, so recordings are not sent to a server for recognition.
“Once the models are downloaded, voice recognition works entirely offline, ensuring your privacy and convenience.” (Initial Setup section)
- Whisper IME's readme says voice recognition works entirely offline once the models are downloaded, so the recognition runs without a network connection.
“Once the models are downloaded, voice recognition works entirely offline, ensuring your privacy and convenience.” (Initial Setup section)
- Whisper IME's readme says it provides an input method editor (keyboard) and can also be selected as system-wide voice input.
“Besides providing an IME, whisper can also be selected as system-wide voice input (RecognitionService)” (Introduction section)
- Whisper IME's readme says voice recognition works entirely offline once the Whisper models have been downloaded.
“Once the models are downloaded, voice recognition works entirely offline, ensuring your privacy and convenience.” (Initial Setup section)