Audio Signal Compression in a Surround Environment Using Wavelet Transform
Miyuki SHIRAI, Shotaro Yamamoto, Tomoyuki Matsumoto, Mikiko Sode · 2024
When a machine emits an abnormal sound, it is often necessary to take measures such as stopping the factory line. Therefore, we have developed a system that detects machine failures using sound. The feature of the proposed system is that it converts voice data into images using wavelet transform, and uses the images as input to determine abnormalities using machine learning. The important thing in this system is the size of the voice data. We would like to compress the voice data to make it easier to send. In this paper, we discuss the compression of voice signals using wavelet transform. We consider reducing the size of voice data without removing abnormal sounds contained in the voice data. We apply several different transformation methods and compare the features of abnormal sounds on the time-frequency plane.