Towards Classification of Technical Sound Events with Deep Learning Models

Abstract

Sounds of machines and mechanical systems contain a lot of information about the observed object and its state. Experienced engineers and technical service staff can often identify or classify a certain technical object with state via its sound. An equivalent automated system with such capabilities is difficult to realise because of noisy unknown surroundings. In this paper, we show an approach to implement the mentioned characteristics with deep learning methods and enhance the power of a technical assistance system.