基于MFCC和CNN的汽车鸣笛声识别算法  

An Recognition Algorithm for Automobile Whistle Based on MFCC and CNN

作　　者：侍艳华 刘菁原[1,2] 卞飞 张小博 雷鸣 SHI Yanhua;LIU Jingyuan;BIAN Fei;ZHANG Xiaobo;LEI Ming(Beijing China Eclectics Intelligent Acoustics Technology Co.,Ltd.,Beijing 100015,China;The No.3 Research Institute of China Electronics Technology Group Corporation,Beijing 100015,China)

机构地区：[1]北京中电慧声科技有限公司,北京100015 [2]中国电子科技集团公司第三研究所,北京100015

出　　处：《电声技术》2020年第5期30-33,共4页Audio Engineering

摘　　要：针对复杂环境下的鸣笛声识别问题,本文设计了一种基于梅尔频率倒谱系数和卷积神经网络的音频分类算法。该算法以梅尔频率倒谱系数特征为深度学习的网络输入,利用卷积神经网络逐层提取声音特征,最终实现汽车鸣笛声的识别。实验后的结果说明,利用该算法后,模型的识别准确率能够达到97.6%以上,能够很好地满足违法鸣笛抓拍系统的应用需求。This paper presents an algorithm of audio classification based on MFCC and CNN,solving the problem that the automobile whistle can’t be detected effectively in the complicated noise circumstances.The algorithm takes MFCC parameters as the network input,then uses the CNN model to obtain the features ofaudioand finally realizes the recognition of the automobile whistle.The experimental results show that the recognition accuracy of the model trained by the algorithm can reach 97.6%,and the algorithm can well meet the application requirements of illegal automobile whistle capturing system.

关 键 词：汽车鸣笛声识别 深度学习 梅尔频率倒谱系数 卷积神经网络 特征提取 

分 类 号：S763[农业科学—森林保护学]