摘要
电子音乐辨识对电子音乐的融合和处理具有十分重要的意义,当前电子音乐辨识方法不能有效描述电子音乐包含的信息,使得电子音乐辨识错误率极高,实际应用价值较低,为了改善当前电子音乐辨识过程中存在的一些难题,以获得更优的电子音乐辨识结果为目标,提出了改进神经网络的电子音乐辨识方法。首先分析当前神经网络在电子音乐辨识建模过程中的缺陷,然后通过引入自适应遗传算法解决神经网络参数优化缺陷,最后将改进的神经网络应用于电子音乐辨识研究中,并与其它电子音乐辨识方法进行了对照测试,结果表明,相对于传统神经网络,改进神经网络的电子音乐辨识精度提升了10%左右,而且电子音乐辨识稳定更好,电子音乐辨识效率也得到了明显的改善,较好地克服了当前电子音乐辨识方法存在的不足,为电子音乐后续处理打下了良好的基础。
Electronic music identification is of great significance to the integration and processing of electronic music.The current electronic music identification methods cannot effectively describe the information contained in electronic music,which makes the electronic music identification error extremely high,and the practical application value is low.For improving the problems that in the curent electronic music identification process.and Aiming at obtaining better electronic music identification results,an electronic music identification model based on improved neural network is proposed.Firstly,the defects of the current neural network in the process of electronic music identification modeling are analyzed,and then an adaptive genetic algorithm is introduced to solve the defects of neural network parameter optimization.Finally,the improved neural network is applied to the research of electronic music identification,and compared with other methods of electronic music identification.The results show that the accuracy of electronic music identification is improved by about 10%,the stability of electronic music identification is better,and the efficiency of electronic music identification is also improved significantly,which overcomes the shortcomings of current electronic music identification methods,and lays a good foundation for the follow-up processing of electronic music.
作者
赵婕
ZHAO Jie(School of information Engineering, Shaanxi Polytechnic Institute, Xianyang 712000, China)
出处
《微型电脑应用》
2021年第6期129-131,135,共4页
Microcomputer Applications
关键词
电子音乐
建模与辨识
神经网络
参数优化
自适应遗传算法
electronic music
modeling and identification
neural network
parameter optimization
adaptive genetic algorithm
