基于高斯低通滤波的音乐节拍提取

摘要

A music beat extraction method based on low-pass Gaussian filter is proposed. And musical energy extraction is applied to calculate the energy envelop. Also denoise is archived by data addition in spatial domain. The quantity of data is reduced also in addition. Then, a low-pass Gaussian filter is formed to depress high frequency components and increase the prominence of low-frequency tempo information.Finally, the beat of music signal is located by peak detection calculation. The experimental results demonstrate that the proposed algorithm is capable of extracting music beat with high accuracy and it is practicable to real-time applications.%提出一种基于高斯低通滤波的节拍提取算法,对音乐信号进行能量分析,确定信号的能量包络.在获得能量包络基础上,采用叠加分析的方法,实现在减少有待分析数据量的同时,消除信号的噪声.然后,应用高斯窗函数构造一个低通滤波器,对压缩后的信号数据进行低通滤波处理,滤除音乐信号中不能决定节拍信息的高频部分.最后,通过峰值检测的方法确定音乐信号的节拍.实验结果表明,本文所采用的方法提取结果准确,且实用性强.