论文部分内容阅读
本文研究了四种分别从语音或气流信号中获取声门气流体积速度波形的方法，即语音单通道，语音──电声门图双通道、气流──电声门图双通道和语音──气流混合式线性预测反滤波声门图方法。因而，将原来仅适用于正常发声的线性预测反滤波声门图方法推广应用到正常与病变发声测试中。合成元音与人体发声实验结果证实了反滤波声门图算法的有效性；提取的声门气流体积速度波形和特征参数反映了嗓音源特性及其变化。
In this paper, four methods for obtaining the volume velocity of glottal gas flow from speech or air flow signals are studied respectively, that is, speech single channel, speech ─ ─ electroacoustic dual channel, airflow ─ ─ electroacoustic diagram dual channel and speech ─ ─ Airflow mixed linear prediction inverse filtering glottic method. Therefore, the original only applies to normal speech linear prediction inverse filtering glottic method to promote the use of normal and pathological vocalization test. Experimental results of synthetic vowel and human vocalizations confirmed the validity of the anti-filtering glottic algorithm. The extracted volume of vowel flow and its characteristic parameters reflect the characteristics of the vocal source and its changes.