The Research On Detecting Method Of Abnormal Response Signal To Loudspeaker Based On Harmonic Distortion

In the production process of the loudspeaker, the fault detection of the abnormal sound is mainly done by manual audition, the standard is difficult to be unified and quantified, and the long listening time can cause great harm to human ear. With the growing demand for electronic products, the number of speakers is also increasing production, listening to the work of the speaker is more difficult, which is the electro-acoustic industry problems to be solved. Therefore, it is an important research topic to develop a system which can detect the abnormal sound by the machine instead of the human ear.This paper analyzes the cause of the abnormal sound of the loudspeaker and the characteristics of the response signal, and studies an on-line detecting method of the abnormal sound of the loudspeaker. The Short-time Fourier Transform of the response signal of the loudspeaker is used to obtain the time-frequency diagram. Through the theory of harmonic distortion, in the time-frequency diagram on the order of harmonic division,the harmonic energy statistics. Then, the feature extractor is designed to extract the characteristics of each harmonic, and find the difference between the qualified speaker and the abnormal speaker. Finally, the classifier is designed to classify qualified and faulty loudspeakers and achieve the purpose of automatic detection.The research results of this paper include:1. Analysis and research on speakers abnormal sound fault and its generation mechanism;2. According to the harmonic distortion mechanism, feature extraction and classification learning, this paper presents a method to detect the abnormal sound of loudspeakers based on harmonic distortion.3. Threshold settings for different types of speakers.4. On-line speaker fault detection system for software and hardware systems integration.