基于非线性预测反滤波声门下嗓音源提取方法的研究

Estimate Methods of Subglottal Voice Source Based on Nonlinear Inverse Filtering Technique

根据嗓音呼吸系统生理和生物物理原理，提出了利用声门下嗓音源研究声带振动特性的新方法，并建立了提取声门下嗓音源的非线性数学模型．在声门上、下声道耦合的条件下，用非线性预测反滤波技术提取了声门下嗓音源，并在不同发声方式条件下运用该模型和人体实验证实了该方法的有效性，克服了线性预测反滤波技术的不足，并消除了声门上、下声道所产生的伪迹．

Physiology ofthe voice respiratory system can be used to abstractthe character ofvocalfold vibration from the subglottal voice source. Anonlinear mathematical model can then be developed to approximated subglottal voice source.Carefulconsideration is given to theinteraction between supra glottaland subglottalvocaltract. Anonlinearinverse filtering technique is then used to suppress the artifacts ofsubglottalvoice source waveforms caused by theinteraction between supraglottal and sub glottal vocaltract.The experimentalresultsin vivo confirm thefeasibility ofthis approach .