DEVELOPMENT OF VOICE ASSISTANT BASED ON NEURAL SYSTEMS

Abstract

The development of a voice assistant based on neural systems is an actual line of research in the field of artificial intelligence. Existing voice assistants were analyzed and their advantages and disadvantages were identified in this paper. The following concepts were considered: natural language processing, and neural networks. Different methods of natural language processing significantly increase the quality of work of many programs, because they can interact with people at a natural level for humans without any problems. Modern voice assistants can help millions of people around the world at the same time, regardless of the time of the request. They are covering more and more spheres every year, from ordering food and taxis and making plans for the day to acting as translators. The types of dialogue systems, the voice assistants` scope of use and their types, recurrent neural networks and data processing models Seq2seq and Transformer, as well as the work of the theoretical component of voice assistants based on neural networks, were considered. Node.js libraries were examined in detail, and an own voice assistant was created and outputs of this system were given. In addition, the possibilities of integrating the voice assistant with other systems and platforms, which expands its functionality and improves interaction with the user, were examined. General theoretical information about voice assistants and their components and principles of work were analysed. An environment for writing a voice assistant was chosen and a practical implementation of writing code for a voice assistant and an example of ready-made code were shown.