Abstract:
In view of the quality problems such as loosening, falling off and other safety risks caused by the hollow drum inside ceramic tile, this paper developed a set of experimental system for the hollow drum rate identification of ceramic tile. The Mel-frequency cepstrum coefficient (MFCC) method was used to extract the characteristic parameters of tile knocking sound, and then the Gaussian mixture model (GMM) method was used to classify and recognize the MFCC characteristic parameters. The test results showed that the method of MFCC and GMM can effectively identify the hollow drum condition of ceramic tile, which had a good application prospect.