摘要
语音信号基音周期检测一直以来都是语音信号处理的关键技术和热点领域。对传统的基音检测方法进行研究分析,提出基于自相关和倒谱法的基音检测改进算法。先将语音信号进行最小均方误差(LMS)自适应滤波和非线性处理进行语音增强,后进行自相关法和倒谱法加权平方运算来检测基音周期。经Matlab实验仿真,该算法在低信噪比环境中能精确检测基音周期,较传统基音检测方法鲁棒性更好、更精确。
Pitch period detection of speech signals has always been a key technique and focus in speech signals processing. According to the study and analysis on traditional pitch detection methods, we propose an ACF and CEP-based improved pitch detection algorithm. First, we carry out the speech enhancement by applying the least mean square （LMS） adaptive filter and nonlinear processing to speech signal, and then run weighted square algorithm of ACF and CEP to detect the pitch period. Demonstrated by the simulation experiments with Matlab, this algorithm can accurately detect the pitch period in the case of low signal-to-noise ratio （ SNR）, and is more robust and precise than traditional pitch detection methods.
出处
《计算机应用与软件》
CSCD
2015年第1期163-166,共4页
Computer Applications and Software
关键词
基音
自相关
倒谱
非线性处理最小均方误差
语音增强
Pitch Autocorrelation function （ACF） Cepstrum （CEP） Nonlinear processing LMS Speech enhancement
