基于小波多分辨分析的基音检测算法及Simulink实现

A pitch detection algorithm based on wavelet multiresolution analysis and its Simulink implementation

为研究小波多分辨分析在信号频带划分中的应用,提出了一种基于小波多分辨分析方法和自相关函数的语音信号基音周期检测算法,在Matlab/Simulink环境下,采用层次化的设计方法和用户自定义函数模块进行了算法系统的设计,以采样频率为8 kHz、采样位数为16 bits的单声道语音信号为处理对象,进行了该算法系统的仿真实验。仿真实验结果验证了该系统的有效性,并取得了良好的检测效果。

The application of wavelet multiresolution analysis to signal frequency band division was discussed. On the basis of it, a speech signal＇s pitch detection algorithm used by the methods of wavelet transform and autocorrelation function（ACF） was presented. The system of the algorithm had been designed in the environment of Matlab/Simulink. And in the process of design, the conception of hiberarehy and user-defined functions were introduced in the system. Finally, that asection of a mono speech signal with 8 kHz sample rate and 16 bits per sample was considered as the object to be processed, the algorithm system had been simulated. Experiment result shows that the algorithm system is valid and the effect of pitch detection is good.