数字助听器中的音频动态范围控制算法研究  

Algorithms for Audio Dynamic Range Control in Digital Hearing Aids

作　　者：丁凯星 梁瑞宇[1] 朱霜霜 DING Kaixing;LIANG Ruiyu;ZHU Shuangshuang(School of Information and Communication Engineering,Nanjing Institute of Technology,Nanjing 211167,China;School of Information Science and Engineering,Southeast University,Nanjing 210096,China)

机构地区：[1]南京工程学院信息与通信工程学院,南京211167 [2]东南大学信息科学工程学院,南京210096

出　　处：《电子器件》2020年第3期652-656,共5页Chinese Journal of Electron Devices

基　　金：江苏省大学生实践创新训练计划项目(201911276021Z)。

摘　　要：本文针对数字助听器提出一种结合语音活动检测的五段动态范围控制算法。传统的动态控制算法只能消除音频信号中携带的噪声,而无法解决播放端的高频噪声问题,而本文采用频域子带能量作为特征,用二维高斯模型进行语音噪声的似然比检验来实时跟踪语音活动可以有效地降低噪声并改善输出信号的失真度。并针对播放端无自动增益控制的情形,在增益曲线的设计时增加了小信号放大功能。最后,通过实验获取了增益曲线中不同功能段的启动时间和释放时间的最优配置,且在输入信噪比为15 dB和20 dB时对输出信噪比分别提升了3 dB和6 dB,对小信号幅度提升了4 dB。A new method of five-stage audio dynamics processor combined with voice activity detection for digital hearing aids is proposed. The traditional dynamic control algorithm can only eliminate the noise carried in the audio signal,but can not solve the high-frequency noise problem at the playback end. The frequency domain sub-band energy is used as the feature,and the two-dimensional Gaussian model is used to test the likelihood ratio of the speech noise. Tracking voice activity in real time can better reduce noise and improve the distortion of the output signal. And for the case that the playback end has no automatic gain control,a small signal amplification function is added in the design of the gain curve. Finally,the optimal configuration of start-up time and release time of different functional segments in the gain curve is obtained through experiments,and the output signal-to-noise ratio is improved by 3 dB and 6 dB respectively when the input signal-to-noise ratio is 15 dB and 20 dB,small signal amplitude is increased by 4 dB.

关 键 词：数字助听器 语音活动检测 动态范围控制 子带能量 高斯模型 

分 类 号：TN912[电子电信—通信与信息系统]