基于音高与端点联合检测的音符识别算法  

Note recognition combining pitch-detection and onset-detection

作　　者：徐鹏进[1] 郭莉[1] 刘书昌[1] 

机构地区：[1]北京邮电大学信息与通信工程学院,北京100876

出　　处：《计算机应用》2011年第A02期172-175,共4页journal of Computer Applications

基　　金：国家科技重大专项(2009ZX03007-001-03)

摘　　要：结合次谐波与谐波比率(SHR)音高提取算法与基于短时能量及过零率的端点检测算法,实现了一种音高、端点联合检测算法。该方法以短时能量及过零率为音符切分的基础,进而利用音高跳变来检测音符的变化。仿真实验表明,采用SHR音高提取算法,该联合检测提高了音符识别的精度,尤其在噪声环境下,其表现优于传统端点检测算法。Using Subharmonic-to-Harmonic Ratio（SHR） algorithm and onset-detection algorithm based on short-time energy and zero-crossing rate,a combined detection algorithm was designed.With short-time energy and zero-crossing rate,the algorithm utilized variation of pitch to realize note segmentation.The simulation results show that the combined algorithm using SHR pitch-detection method improves the accuracy of note recognition,and it＇s more effective than traditional onset-detection algorithm in noisy environment.

关 键 词：次谐波与谐波比率 短时能量 短时过零率 端点检测 基音检测 音符分割 

分 类 号：TP391.42[自动化与计算机技术—计算机应用技术]