This project provides a voice keyboard using Vosk. It is based on https://github.com/Felicis/vosk-android-demo.
This app requires a Vosk model to run. You can use the built-in downloader or download manually from https://alphacephei.com/vosk/models, and then import into the app.
Sayboard uses for the following permissions:
- INTERNET - to download the models. This permission is not required, and can be revoked through the device settings in compatible ROMs. Sayboard does not, and never will, send any of your information over the internet!.
- RECORD_AUDIO - well, this is a voice keyboard after all.
- POST_NOTIFICATIONS - to show download and import progress.
- FOREGROUND_SERVICE - to download and import models in the background, and for the RecognitionService.
- FOREGROUND_SERVICE_MICROPHONE - for the RecognitionService.
- FOREGROUND_SERVICE_SPECIAL_USE - for the download service.
- QUERY_ALL_PACKAGES - do to bug in Android, this permission is required for a speech RecognitionService to work properly (see Kaljurand/K6nele-service#9).
Screenshot of Sayboard:
screen.mp4
Note: prior to commit Multiple language support!,
this repository had another gradle model named models. As it is large and not needed anymore, I removed it from git history.
If you want to use any earlier commits, you can find it here.