摘要
基音周期检测一直是音频处理领域的研究热点,基音周期的精确检测实际上是一件比较困难的事情。提出了一种LPC残差与SCMDSF相结合的基音周期检测,该算法的特点在于着重对被处理的语音进行滤波预处理,提取语音信号的LPC残差,消除了声道响应信息,对求出的语音残差信号做SCMDSF计算,并求出语音的基音周期。实验表明,在噪声环境下这种处理方法能够比较准确的提取基音周期。
Pitch period detection is always one of the focuses in voice frequency process, and the precise detection of the pitch period detection is actually a rather difficult matter. This paper proposes a kind of pitch period detection with the combination of residu- al of LPC and SCMDSF. The characteristic of this algorithm lies in the emphatic filtering pretreatment of the speech sounds being processed, the collection of residual of LPC of sound signals, the elimination of the response information in the vocal tract, the SCMDSF calculation for the speech sound signals that have been worked out, and working out the pitch period. Experiments indicate that this process method can collect pitch period comparatively precisely in the noise environment.
出处
《微计算机信息》
2010年第6期25-26,29,共3页
Control & Automation
基金
基金申请人:于洪志
项目名称:基于语音
嗓音和呼吸信号的藏语拉萨话韵律模型研究
基金颁发部门:国家自然科学基金委(60773052)