数字助听器中多通道响度补偿算法的研究

Research on Multi-channel Loudness Compensation in Digital Hearing Aids

目前数字助听器中用的较多的响度补偿方法,通常是对语音的高频部分进行高增益,但由于听损患者在高频段的听觉范围比较小,直接对高频部分进行高增益容易超出听损患者的听觉范围。文中引进了宽动态压缩技术,首先将整个语音的动态范围按一定比例均匀压缩到患者的残余听力中,然后将频带非等宽划分,再在不同频段进行响度补偿和增益控制,最后对信号进行重构,并利用小波阈值去噪技术去除部分噪声。实验结果表明,该方法能够使补偿后语音的响度完全映射到听损患者的听觉范围内,从而有效提高患者的听力水平,特别对高频部分语音的辨识,具有较好的效果。

Loudness compensation algorithm is a key technique in digital hearing aid. At present, digital hearing aid in multi-channel loudness compensation method,is usually high gain for the high frequency part of speech. However,because of the relatively small range of hearing loss in the high frequency range, the high gain of the high frequency part is easy to exceed the range of hearing loss. Wide dy- namic compression technique is introduced in this paper. First the dynamic range of the speech according to a certain proportion is uni- formly compressed to patients with residual hearing, then the band is partitioned with non equal width, and loudness compensation and gain control is carded out in different frequency band. Finally,the signal is reconstructed, removing partial noise by wavelet threshold de- noising technology. Experiment shows that the method makes the loudness of speech after compensation completely mapped to range of hearing loss in patients,effectively improving the patient＇ s hearing level,especially on speech identification of high frequency part with good effect.