基于优化能量值门限和增强倍频效应的抗噪基音检测算法
|
摘要
针对基音检测中的倍频、半频误差问题,提出了一种新的基于优化能量值门限和增强倍频效应的抗噪基音检测改进算法.该算法首先通过修改频谱能量值门限来优化半频误差,然后通过音频特有的倍频效应,取相邻的基频与泛频或泛频与泛频做差,从而得到基音频率,来优化倍频误差.结果表明:所改进的基音检测算法不仅保留了对共振峰的高容忍度,而且大幅提高了在低频区域和高频区域的音频识别能力;该方法比传统基音检测方法不论是在有噪音还是无噪音、高信噪比还是低信噪比的情况下都更加优秀,尤其体现在100Hz以下的低频区域和800Hz以上的高频区域.
In this study,a novel algorithm is proposed based on optimized energy threshold and enhanced frequency doubling effect to solve frequency multiplication and half frequency error in pitch detection.The algorithm first optimizes the half frequency error by modifying the spectrum energy threshold,and then makes the difference between the adjacent fundamental frequency and the pan frequency or the pan frequency and the pan frequency by the unique frequency doubling effect of the audio,so as to get the pitch frequency and optimize the frequency doubling error.Experiments show that the improved pitch detection algorithm not only preserves the high tolerance of formants,but also greatly improves the audio recognition ability in low frequency regions and high frequency regions.When compared with traditional pitch detection methods,this method is superior to noise or no noise in high SNR or low SNR,especially in low frequency regions below 100 Hz and high frequency regions above 800 Hz.
In this study,a novel algorithm is proposed based on optimized energy threshold and enhanced frequency doubling effect to solve frequency multiplication and half frequency error in pitch detection.The algorithm first optimizes the half frequency error by modifying the spectrum energy threshold,and then makes the difference between the adjacent fundamental frequency and the pan frequency or the pan frequency and the pan frequency by the unique frequency doubling effect of the audio,so as to get the pitch frequency and optimize the frequency doubling error.Experiments show that the improved pitch detection algorithm not only preserves the high tolerance of formants,but also greatly improves the audio recognition ability in low frequency regions and high frequency regions.When compared with traditional pitch detection methods,this method is superior to noise or no noise in high SNR or low SNR,especially in low frequency regions below 100 Hz and high frequency regions above 800 Hz.
引文
[1]孙朝平.关于十二平均律音分与频率的换算[J].乐器,2017,3:22-23.
[2]DING H,SOON Y,YEO CK.A DCT-based speech enhancement system with pitch synchronous analysis[J].Speech and Language Processing,2011,19(8):2614-2623.
[3]AHMADI S,SPANIAS A S.Cepstrum-based pitch detection using a new statistical V/UV classification algorithm[J].Speech and Audio Processing,1999,7(3):333-338.
[4]JIN Z,WANG D L.HMM-based multipitch tracking for noisy and reverberant speech[J].Speech and Language Processing,2011,19(5):1091-1102.
[5]陈盼弟,黄华,何凌.基于自相关和倒谱法的基音检测改进算法[J].计算机应用与软件,2015,32(1):163-166.
[6]王丽.基于MATLAB的自相关函数基音检测的优化[J].电脑知识与技术,2009,5(36):10611-10612.
[7]梁颂朗.基于去噪的ACF_CEP基音检测算法[J].中国科技信息,2008,12:37-41.
[8]LYUDMILA SUKHOSTAT,YADIGAR IMAMVERDIYEV.A comparative analysis of pitch detection methods under the influence of different noise conditions[J].Journal of Voice,2015,29(4):410-417.
[9]刘卫.人声浊音基频测量谱分析算法[J].现代电子技术,2008,4:113-120.
[10]GHULAM MUHAMMAD.Noise-robust pitch detection using auto-correlation function with enhancements[J].Journal of King Saud University,2010,22:13-28.
[11]WU KEBIN,ZHANG DAVID,LU GUANGMING.iPEEH:improving pitch estimation by enhancing harmonics[J].Expert Systems With Applications,2016,64:317-329.
[12]骆娇艳,孙祥娥.基于MATLAB的基音检测分析[J].电脑知识与技术,2014,10(18):4292-4295.
[13]王芸,黄华,何凌,等.基音周期检测算法研究[M].北京:电子工业出版社,2005:377-381.
[14]王浩军.基于窗函数的数字FIR滤波器设计[J].舰船电子工程,2017,8:169-171.
[15]SANJAY KUMAR,KULBIR SINGH,RAJIV SAXENA.Analysis of dirichlet and generalized“hamming”window functions in the fractional fourier transform domains[J].Signal Processing,2011,91(3):600-606.
[16]FABRCIO LOPES SANCHEZ.Wavelet-based cepstrum calculation[J].Journal of Computational and Applied Mathematics,2009,227(2):288-293.
[17]周治国,王毅,张华良.高频采样下基于DFT的配电网相量测量算法研究[J].电力系统保护与控制,2017,19:101-108.
[18]杨鑫,王辉,李勇,等.基于频域融合的计算机制半周视彩色彩虹全息[J].中国激光,2017,1:1-14.
[19]张婷,何凌,黄华,等.基于小波及能量熵的带噪语音端点检测算法[J].计算机工程与设计,2013,34(4):1331-1335.
[20]李艳坤,王涵,曾祥超,等.微量元素的ROC曲线和T检验对肺癌的诊断价值[J].计算机与应用化学,2017,5:376-378.
[21]LUIGI ATZORI,ANTONIO IERA,GIACOMO MORABITO.The Internet of things:a survey[J].Computer Networks,2010,54(15):2787-2805.
[22]宋黎明,李明,颜永红.谐波显著度的基频提取方法[J].声学学报,2015,40(2):294-299.
[2]DING H,SOON Y,YEO CK.A DCT-based speech enhancement system with pitch synchronous analysis[J].Speech and Language Processing,2011,19(8):2614-2623.
[3]AHMADI S,SPANIAS A S.Cepstrum-based pitch detection using a new statistical V/UV classification algorithm[J].Speech and Audio Processing,1999,7(3):333-338.
[4]JIN Z,WANG D L.HMM-based multipitch tracking for noisy and reverberant speech[J].Speech and Language Processing,2011,19(5):1091-1102.
[5]陈盼弟,黄华,何凌.基于自相关和倒谱法的基音检测改进算法[J].计算机应用与软件,2015,32(1):163-166.
[6]王丽.基于MATLAB的自相关函数基音检测的优化[J].电脑知识与技术,2009,5(36):10611-10612.
[7]梁颂朗.基于去噪的ACF_CEP基音检测算法[J].中国科技信息,2008,12:37-41.
[8]LYUDMILA SUKHOSTAT,YADIGAR IMAMVERDIYEV.A comparative analysis of pitch detection methods under the influence of different noise conditions[J].Journal of Voice,2015,29(4):410-417.
[9]刘卫.人声浊音基频测量谱分析算法[J].现代电子技术,2008,4:113-120.
[10]GHULAM MUHAMMAD.Noise-robust pitch detection using auto-correlation function with enhancements[J].Journal of King Saud University,2010,22:13-28.
[11]WU KEBIN,ZHANG DAVID,LU GUANGMING.iPEEH:improving pitch estimation by enhancing harmonics[J].Expert Systems With Applications,2016,64:317-329.
[12]骆娇艳,孙祥娥.基于MATLAB的基音检测分析[J].电脑知识与技术,2014,10(18):4292-4295.
[13]王芸,黄华,何凌,等.基音周期检测算法研究[M].北京:电子工业出版社,2005:377-381.
[14]王浩军.基于窗函数的数字FIR滤波器设计[J].舰船电子工程,2017,8:169-171.
[15]SANJAY KUMAR,KULBIR SINGH,RAJIV SAXENA.Analysis of dirichlet and generalized“hamming”window functions in the fractional fourier transform domains[J].Signal Processing,2011,91(3):600-606.
[16]FABRCIO LOPES SANCHEZ.Wavelet-based cepstrum calculation[J].Journal of Computational and Applied Mathematics,2009,227(2):288-293.
[17]周治国,王毅,张华良.高频采样下基于DFT的配电网相量测量算法研究[J].电力系统保护与控制,2017,19:101-108.
[18]杨鑫,王辉,李勇,等.基于频域融合的计算机制半周视彩色彩虹全息[J].中国激光,2017,1:1-14.
[19]张婷,何凌,黄华,等.基于小波及能量熵的带噪语音端点检测算法[J].计算机工程与设计,2013,34(4):1331-1335.
[20]李艳坤,王涵,曾祥超,等.微量元素的ROC曲线和T检验对肺癌的诊断价值[J].计算机与应用化学,2017,5:376-378.
[21]LUIGI ATZORI,ANTONIO IERA,GIACOMO MORABITO.The Internet of things:a survey[J].Computer Networks,2010,54(15):2787-2805.
[22]宋黎明,李明,颜永红.谐波显著度的基频提取方法[J].声学学报,2015,40(2):294-299.