一种基于短时傅里叶变换的扬声器异常音检测方法技术

本发明专利技术提出了一种基于短时傅里叶变换的扬声器异常音检测方法。本发明专利技术采用扫频信号激励被测扬声器，通过采集被测扬声器的声响应信号并对该信号进行短时傅里叶变换，从而获得该信号所对应的时频图。根据扬声器异常音信号的特点对该时频图进行分割，然后量化分割后的各个区域与标准时频图对应区域的差异程度，并将得到的差异程度进行显示方式转换，构造出异常音曲线。最后，将该异常音曲线与设定的门限进行比较，从而判断扬声器是否存在异常音。方法中所提的门限是由良品时频图与标准时频图的最大差异程度的M倍所构造出的异常音曲线。

A method of speaker abnormal tone detection based on short time Fourier transform

The invention provides a loudspeaker abnormal sound detection method based on short time Fourier transform. The invention uses the sweep frequency signal to excite the tested loudspeaker, and acquires the sound response signal of the tested loudspeaker and carries out a short time Fourier transform to the signal so as to obtain a time frequency diagram corresponding to the signal. According to the characteristics of abnormal sound signal to the speaker time-frequency map segmentation, and each region with the standard quantization segmented time-frequency map corresponding to regional differences, and the differences in the degree of the display mode conversion, construct the abnormal sound curve. Finally, the abnormal tone curve is compared with the set threshold to determine if there is an abnormal sound in the speaker. The threshold proposed in the method is the abnormal tone curve constructed by the M difference of the maximum difference between the time-frequency spectrum of the good product and the standard time-frequency map.

【技术实现步骤摘要】

【技术保护点】
一种基于短时傅里叶变换的扬声器异常音检测方法，其特征在于该方法包括以下步骤：首先，采用扫频信号激励被测扬声器，同时采集被测扬声器的声响应信号并对该响应信号做短时傅里叶变化得到其对应的时频图；之后，根据扬声器异常音信号的特点对该时频图进行分割并量化分割后的各个区域与标准时频图的差异程度，通过计算机将得到的差异程度进行显示方式转换，构造出异常音曲线；最后，将获得的异常音曲线与设定的门限进行比较，从而判断扬声器是否存在异常音；上述所提及的门限是已知合格样品的异常音曲线的M倍；M的取值范围一般是1.3~1.5。

【技术特征摘要】

【专利技术属性】
技术研发人员：李宏斌，温周斌，徐楚林，计敏君，金晶，
申请(专利权)人：浙江中科电声研发中心，嘉善恩益迪电声技术服务有限公司，
类型：发明
国别省市：