数字助听器中回声消除算法的研究及改进

发布时间：2018-04-25 15:16

本文选题：数字助听器 + 回声消除　；参考：《南京邮电大学》2017年硕士论文

【摘要】：对听力受损患者来说,数字助听器是其生活中不可缺少的工具,使用助听器可以有效改善他们的听力水平。回声反馈现象是数字助听器使用过程中一种常见问题,严重影响患者的使用效果。本文重点研究回声反馈消除技术,本文内容如下:(1)分析了语音的发声机理和人类的听觉模型;研究了有关语音的基础理论;重点介绍了在数字助听器中使用的几项关键技术,包括回声反馈消除、响度补偿、声源定位和语音增强等,其中着重关注了回声消除算法的研究现状。(2)传统的NLMS算法在处理具有强相关的语音时,算法效果会出现严重恶化的情况,本文将重点研究仿射投影算法(Affine Projection Algorithm)在回声消除中的应用。算法迭代过程中步长的大小直接影响收敛速度与稳态失调,本文介绍一种新算法,该算法基于离散余弦变换的变步长仿射投影算法,与传统的仿射投影算法对比,该算法在实现较快收敛速度的同时失调也更加低,大大提高了算法的使用效果。(3)仿射投影算法因为重复使用语音数据,造成该算法的算法复杂度比较大,影响数字助听器的实时性。本文在变步长投影算法的基础之上加入变投影阶数算法,在迭代过程中不断降低算法的投影阶数,从而降低算法计算复杂度。仿真测试表明,该算法在保证收敛速度和失调量的同时极大降低了算法复杂度,保证了算法在使用过程中的实时性。
[Abstract]:Digital hearing aid is an indispensable tool in the life of hearing impaired patients, and the use of hearing aids can effectively improve their hearing level. Echo feedback is a common problem in the use of digital hearing aids. This paper focuses on echo feedback cancellation technology, which is as follows: 1) analyzes the sound mechanism of speech and human auditory model, studies the basic theory of speech, and introduces several key technologies used in digital hearing aid. Including echo feedback cancellation, loudness compensation, sound source location and speech enhancement, among which, emphasis is placed on the research status of echo cancellation algorithms. In this paper, the affine Projection algorithm will be applied to echo cancellation. In this paper, a new algorithm is introduced, which is based on discrete cosine transform variable step size affine projection algorithm, which is compared with the traditional affine projection algorithm. The algorithm achieves faster convergence speed and lower misalignment, which greatly improves the performance of the algorithm. The affine projection algorithm is more complex because of the repeated use of speech data. It affects the real time of digital hearing aid. In this paper, the variable projection order algorithm is added to the variable step size projection algorithm, and the projection order of the algorithm is reduced in the iterative process, thus reducing the computational complexity of the algorithm. The simulation results show that the algorithm not only guarantees the convergence speed and misalignment but also greatly reduces the complexity of the algorithm and ensures the real-time performance of the algorithm.
【学位授予单位】：南京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN912.3;R764.5

【参考文献】