This article deals with the design of a system for monitoring gearbox vibrations, which is one of the initial phases of research within a dissertation. This study examines the possibility of diagnosing and predicting failures in military vehicles using machine learning to increase the operational reliability of military equipment and enable the early detection of impending failures. For the purposes of machine learning, it is necessary to obtain a large amount of input data so that it can subsequently be compared with a state in which a potential defect can be observed. Essentially, there are two options for obtaining the input dataset: using a ready-made dataset or acquiring data from a perfect state. A report on this method of data acquisition is the subject of this article. This article presents a real-time vibration monitoring solution that is beneficial for the aforementioned purposes. During the development of this report, efforts were made to utilize commercially available devices to create a cost-effective solution that could be used to acquire input data and subsequently compare it.