中文摘要：

听觉场景匹配是当前智能数字助听器研究的热点问题,目前匹配精度大都较为理想,然而过高的计算量还是影响了硬件处理实时性的实现。此外,各类研究中所涉及的听觉场景类型和数量与人们实际接触较多的听觉场景也有差别。为此,首先对日常听觉场景类型进行分析统计,在此基础上提出了一种简单高效的智能数字助听器听觉场景匹配方法,以临界带能量比例为特征,以最小距离为分类算法,计算量较小、精度高,更适合数字助听器对计算量的苛刻要求。最后,通过主观测听,证明该方法与人耳的识别效果类似。

英文摘要：

Recently,the recognition of acoustic scenes has been a hot topic in the field of intelligent digital hearing aids.Though the recognition precision is high in most current studies,the huge computational load remains a big problem that influences the implementation of hardware real-time processing.In addition,there exist differences between researches and peoples’ daily lives in classifying and quantifying acoustic scenes.In this study,the analysis and statistics to daily acoustic scenes are addressed firstly,and then an efficient method of recognizing acoustic scenes is presented,which uses the critical band ratio as features and the minimum distance as classifier.Compared with other methods,this method is more suitable for digital hearing aids with limited resource because of the lower load and the high precision.Finally,it is shown from subjective listening tests that the proposed method has similar effects compared with auditory perception.