基于调频调幅的耳鸣匹配声源合成及多样化

耳鸣掩蔽疗法是临床治疗耳鸣的常用方法.国内外传统的耳鸣治疗仪普遍存在耳鸣匹配声源不足且匹配效果差的缺陷.针对该问题,本文基于调频调幅语音模型,结合能量分离算法,提出一种分离合成及多样化最常见的耳鸣声源——蝉鸣声的方法.实验表明,该方法取得了较好的效果,丰富了蝉鸣声匹配声源,且计算复杂度低;能够拓展到实现其他耳鸣匹配声源的合成及其多样化,具有一定的医用参考价值.

Synthesis and diversification of tinnitus matching sound based on Amplitude-Frequency modulation

Masking therapy is a commonly used clinical treatment to cure tinnitus. The domestic and foreign traditional therapeutic instrument generally exists the defects of lacking in masking sound sources and poorly matching effect. Thus, this paper proposes a method to synthesize and diversify the most common tinnitus sound source cicada sound using AM FM speech model and energy separation algorithm. Experimental results indicate that this method can successfully synthesize diverse cicada sound with low computational complexity. In addition, this method can be applied to synthesize other masking sound sources, which makes it a valuable medical reference.