Dialogue System based on Reinforcement Learning in Smart Home Application

Abstract

This research discusses how to interact with a smart home using speech recognition and a touchscreen to control electronic devices. Google Speech Cloud API use to process speech-to-text and text-to-speech. The system is built in a mobile-based application using a touchscreen as remote control and speech to control the electronic devices. This mobile application is made using the Flutter framework. We use natural language understanding (NLU) in speech processing to determine the intent. The learning process in a dialogue system is based on reinforcement learning. Interaction through the touch screen on the mobile application performs well, while the dialogue system based on reinforcement learning accuracy rate is 83.33%.