针对传统智能照明安全控制系统中由于声纹特征提取单一而影响识别准确率的问题,文章提出融合梅尔频率倒谱系数(Mel Frequency Cepstrum Coefficient,MFCC)与伽马通频率倒谱系数(Gammatone Frequency Cepstral Coefficients,GFCC)特征参数,采用费希尔准则对融合后的特征参数进行降维,利用卷积神经网络(Convolutional Neural Networks,CNN)构建智能照明系统声纹识别模型,通过搭建实验环境对改进后的方式进行测试.实验结果表明,改进后的方式能够有效提升系统声纹识别的准确性,提升系统整体的安全性.
Security Control System of Intelligent Lighting System Based on Voiceprint Recognition
Aiming at the problem that the recognition accuracy of traditional intelligent lighting safety control system is affected by single voicing feature extraction,this paper proposes a fusion of Mel Frequency Cepstrum Coefficient,the feature parameters of MFCC and Gammatone Frequency Cepstral Coefficients(GFCC)are reduced dimensionally by using Fisher criterion.Convolutional Neural Networks(CNN)are used to construct the voicing recognition model of intelligent lighting system,and the improved method is tested by setting up an experimental environment.The experimental results show that the improved method can effectively improve the accuracy of voiceprint recognition and improve the overall security of the system.
acoustic pattern recognitionintelligent lightingsecurity control
NETL
NSTL
万方数据
