基于SVM的录音设备分类研究

Recording Equipment Classification Study Based on SVM

为解决音频取证中私录音频由何种录音设备所录的问题,针对不同设备所采用的压缩算法不同,就会导致录音信号中蕴含着区别于其他录音设备的个性特征,本文从压缩算法出发,提出了一种基于SVM对录音设备的分类方法。首先获取不同录音格式的音频,然后针对音频分别用MATLAB对其求改进MFCC倒谱参数,接着选定测试集和训练集,使用交叉验证方法得到倒谱数据的最佳参数,之后用训练集对SVM进行训练,再用得到的模型来预测测试集的分类标签。通过仿真与实验,结果表明,该方法能够较好的区分不同压缩算法下的音频特性,平均识别率达97%。

To solve the problem of which kind of recording equipment is used for private audio recorded in audio forensic, the article presents a classification method for recording equipment based on SVM embarking from the compression algorithm, which is based on the fact that the recorded signals from different devices with different compression algorithms contain personality characteristics different from other recording devices.Audios in different format are collected at first.Then its improved MFCCs are extracted respectively by MATLAB and testing and training sets are selected. Then Cross Validation method is used to get the optimal parameters of cepstrum data.The SVM is trained with the training set and the classification label of the testing set is predicted with the model obtained.The simulation and test results show that the method can distinguish audio features among different compression algorithms better,and the average recognition rate is 97%.