Ciabata is a simple (yet customizable) fitness timer for iOS that follows the Tabata method. It uses the Spokestack libary to enable hands-free control via voice as well as user-supplied custom TTS feedback.
Ciabata is a simple Android app used to demonstrate a basic integration with the Spokestack voice control library. It follows the Tabata method, alternating cycles of high-intensity work and rest over a period of two minutes (all times are configurable in the TimerSettings class) and allowing the user to control the timer with their voice.
First, download the TensorFlow Lite wakeword models:
By default, the app uses Android's built-in SpeechRecognizer for ASR, but the Android emulator doesn't play well with this component, so you'll either need to test on a real device or use a different ASR to test voice features.
To use Google ASR, you'll need a google.properties file (in the Java Property file format) in the root of the project with a single property, CREDENTIALS, that contains Google API credentials in a stringified version of the JSON format described here. Including a JSON string in Java's property format is admittedly inelegant, so feel free to change the app's Gradle file to load these credentials some other way; it's done this way for pure expedience.