文章摘要

文章信息/Info

Title:: A new method of sound synthesis based on phase vocoder for multi-track personalized tinnitus masking therapy

Author(s):: CHEN Yanan; HE Peiyu; XIE Ziqi; LI Yongkang; FANG Ancheng; PAN Fan; （College of Electronics and Information Engineering, Sichuan University, Chengdu 610065,China）

Keywords:: Tinnitus sound therapy; Phase vocoder; Masking therapy; Natural sound; Frequency matching; Personalized

摘要:: 针对传统掩蔽治疗音在耳鸣声治疗中有效性和可听性不足的问题，本研究提出了一种基于相位声码器（phase vocoder, PV）和渐变混合算法的个性化多轨耳鸣掩蔽治疗音（multi-track personalized tinnitus masking therapy sound，MT-PTMTS）合成方法。本研究以生物自然音库和环境背景自然音库为基础，利用PV进行音调变换，同时使用迭代函数系统（iterated function system，IFS）分形映射算法合成能够匹配患者耳鸣频段的时长可控且相似不重复的个性化生物自然音，并加入经渐变混合算法拼接而成的环境背景自然音。实验结果表明，本研究合成的治疗音在频段上可有效匹配患者的耳鸣频段，且符合1/f波动；此外，治疗音输入耳鸣模型能有效抑制对应频段耳鸣。试听结果表明，治疗音怡人且自然,对耳鸣声治疗具有一定的参考价值。

Abstract:: To solve the effectiveness and audibility of traditional masking therapy in tinnitus treatment， we proposed a multi-track personalized tinnitus masking therapy sound(MT-PTMTS)synthesis method based on phase vocoder（PV） and gradient blending algorithm. Based on the biological natural sound database, and the environmental background natural sound database. PV was used for tone transformation.The iterated function system（IFS） fractal mapping algorithm was used to synthesize personalized biological natural sounds that could match the frequency band of patients′ tinnitus with controllable duration and were similar without repetition. Finally, the environmental background natural sounds that were spliced by gradual mixing algorithm were added. The experimental results showed that MT-PTMTS could effectively match the tinnitus frequency band of patients, and it conformed to the 1/f fluctuation. In addition, MT-PTMTS could effectively suppress the corresponding frequency band tinnitus after input to the tinnitus model. The results show that the therapeutic sound is pleasant and natural, and has certain reference value for tinnitus treatment.