机构地区: 桂林航天工业高等专科学校
出 处: 《测控技术》 2010年第7期16-19,23,共5页
摘 要: 根据语音信号产生的机理,综合考虑基音频率检测的准确性与高效性,提出了一种基于下采样的基音检测方法。该方法以传统的自相关函数法为基础,结合线性预测逆滤波及下采样技术,有效地提高了抗噪性,降低了计算复杂度。通过Matlab软件对计算结果进行仿真分析,证明新算法计算简便、准确,基音频率平滑性好,算法鲁棒性好,适用于实时语音信号的基音检测。 According to the mechanism of speech signal,synthesized with accuracy and efficiency of base fundamental frequence,a pitch detection based on down-sampling is proposed.Based on the classical autocorrelation function,combined with LP inverse filter and down-sampling,antinoise performance is improved,and the calculated complication is lessened.By Matlab software simulation of the algorithm,compare with the classical method of autocorrelation function and LPC-autocorrelation function,the simulate results indicate this algorithm has simplicity,validity,pitch smoothing,the good accuracy and robustness,and is proper to calculate pitch frequency in real-time speech signal process.