We use Vosk for Speechrecognition. https://github.com/alphacep/vosk-api/

Vosk is used when activated by the user (single tap on EyeAIVision, clicking the button once in the App).

There is only one class for the implementation (VoskModel.kt).