民族乐器识别方法研究与实现

Research and Realization of Identification Methods on Ethnic Musical Instruments

中国的民族乐器种类繁多,是不可多得的非物质文化遗产。民族乐器的识别分类,目的是让世界上更多的人能够了解少数民族乐器,使少数民族的音乐源远流长。通过MATLAB对民族乐曲进行分段、分帧、加窗等预处理,求得相应的MFCC特征向量,应用LVQ神经网络实现民族乐器的识别。通过3500组训练集数据、1500组测试集数据进行识别分类,识别率最高可达97.5%,说明本文提出的方法对民族乐器识别具有一定的参考价值。

There are many kinds of ethnic musical instruments in China,which are rare intangible cultural heritage.The purpose of the identification and classification for ethnic instruments is letting more people in the world know about ethnic instruments and making the history of ethnic music continue.The ethnic music is segmented,framed and added windows using MATLAB.The corresponding MFCC feature is obtained and the ethnic musical instruments are identified using LVQ neural network.Through 3500 sets of training data and 1500 sets of test data,the recognition rate can reach up to 97.5%,which indicates that the method proposed in this paper has certain value for the identification of ethnic musical instruments.