Anomalous Sound Analysis Technology for Factories That Reduces Network Load

When anomalous sounds occur from machinery, it is almost always necessary to take measures such as stopping the factory line, which often results in significant losses. For this reason, anomalous sounds analysis is important in factories. In recent years, anomalous sound analysis using machine learning has been proposed. When using supervised learning, it is said that it is possible to make nearly 100% accurate judgments. There are many machines in a factory, and it is normal for there to be multiple factories. This means that a large amount of data needs to be collected and learning needs to continue. Collecting a large amount of voice data increases the network load. This paper describes the results of an investigation into compression methods for compressing and transmitting voice data. The conditions for compression are that machine learning can be performed using compressed data and that anomalous sounds can be identified by using machine learning.