基于听觉仿生模型的乐器识别

Musical instrument recognition based on the bionic auditory model

提出了一个用于乐器识别的听觉仿生系统.该系统基于人类听觉系统中对声源识别起关键作用的耳蜗基底膜、内毛细胞、耳蜗核等部位的生理学功能设计出听觉仿生系统,并对乐器特征进行了提取,进而建立模拟听皮层功能的自组织特征映射神经网络,对构建的由7种乐器243个独奏乐曲样本组成的乐器数据库进行分类识别,乐器识别正确率在75%以上.

This paper presents a bionic auditory system for musical instrument recognition. Based on the physiological structures of the human auditory system that are essential to sound source recognition, such as the basilar membrane,inner hair cells and cochlear nucleus, this system is designed to extract the features of the musical instruments, and a self-organizing mapping neural network （SOMNN） based on the function of auditory cortex is established to classify the large solo database, which consists of 243 acoustic and synthetic solo tones of seven different instruments. The instruments are recognized with an average success rate of over 75%. The result indicates that this bionic auditory system is high efficiency and high accuracy in musical instrument recognition.