Human-Computer Interaction Through Voice Commands Recognition

Abstract

Despite intensive research dedicated to human machine communication, until recently only a reduced number of systems adopted vocal commands to interact with computers. With the development of deep convolutional neural networks (DCNNs) architectures promising results have been obtained in speech recognition and natural language processing, allowing people to engage in simple question-answer dialog with robots. In that regard, the goal of this paper is to introduce a novel interface designed to facilitate the human-machine interaction (HMI). The framework starts by detecting and recognizing speech messages, transforming them into spoken commands. The voice controls are further converted into operating system instructions designed to automate everyday human tasks. The experimental evaluation performed on English spoken language dataset, validates the proposed architecture with average accuracy scores superior to 89%. In addition, we have compared different DCNNs architectures and identified which model has the best performance.