医用蝉鸣声多样化合成的一种新方法  

A new method of diversification synthesis of the sound of medical tinnitus

作　　者：吕多玉[1] 何培宇[1] 潘帆[1] 

机构地区：[1]四川大学电子信息学院,成都610065

出　　处：《四川大学学报（自然科学版）》2015年第3期567-572,共6页Journal of Sichuan University(Natural Science Edition)

基　　金：四川省科技支撑项目资助(2011SZ0123);四川省科技支撑项目资助(2013GZ1043)

摘　　要：耳鸣掩蔽疗法是根据耳鸣的频谱和响度来调节掩蔽声,使之与耳鸣感知类型相匹配,以达到掩蔽耳鸣的目的.蝉鸣声是最为常见的耳鸣感知类型,但不同患者的耳鸣感知类型不一样,因此研究蝉鸣声的多样化合成方法显得至关重要.针对耳鸣匹配声源不够丰富的问题,本文将近年来迅速发展起来的希尔伯特-黄变换法和非线性调幅-调频模型(AM-FM)相结合的方法来合成蝉鸣声.首先,利用总体平均经验模态分解法的自适应带通滤波特性,将不同共振峰分离出来;然后通过归一化希尔伯特算法计算出各自的瞬时频率和瞬时幅度;最后利用AM-FM合成蝉鸣声.实验结果表明,该方法合成效果较理想,具有一定的医用参考价值.Tinnitus masking therapy is adjusting masking sound based on the spectrum and loudness of tinnitus, so as to making it matches with type of tinnitus perception, reaching the goal of masking tinni- tus. Cicadas is the most common type of tinnitus perception, but different patients＇ perception of tinni- tus is not the same, so researching the diversed analysis methods of cicadas is critical. To address the problem of not enough of tinnitus matched sound source, we synthesis cicadas sound making use of the Hilbert-Huang Transform developed rapidly in recent years cobined with amplitude modulation frequen- cy modulation model （AM-FM）. First, taking advantage of the adaptive band-pass filter characteristic of the overall average empirical mode decomposition method, after, we separated different formant and cal- culate their instantaneous frequency and amplitude by normalized Hilbert algorithm, then synthesised ci- cadas sound combined with AM-FM. Experimental results indicate that this method has an ideal effect, which makes it a valuable medical reference.

关 键 词：经验模态分解 调幅-调频模型 耳鸣声源 掩蔽疗法 

分 类 号：TN911[电子电信—通信与信息系统]