The main goal of this final work is to get acquainted with the way of training the machine learning model which will be used to recognize short voice commands. More precisely it is necessary to train a model on your own dataset and to show its main recognition functionality through developed application. The first part of the paper describes the area of machine learning and neural networks as well as a way of working with audio recordings. The following chapter is a detailed description of the entire model creation process. In order to demonstrate the work of the model, a mobile application was created using Java Programming Language and Android Studio development environment. The application retrieves all MP3 audio files from the mobile device storage and creates a list of items that can be run. With the integration of the mentioned model, users are allowed to control the application by voice commands. The paper contains a display of all application screens, a description of the recording process and audio processing in the Java programming language, as well as the most important software snippets.