基于支持向量机的咳嗽自动识别

Automatic Recognition of Cough Based on Support Vector Machine

为了进一步改善咳嗽自动识别的效果,本文以支持向量机作为咳嗽识别的分类模型,详细介绍样本采集、MFCC特征参数提取和支持向量机咳嗽识别的实现过程,并与隐马尔可夫模型和动态时间规划的识别结果及运行时间进行比较。实验结果表明在识别率方面,当训练样本集较大时,支持向量机与隐马尔可夫模型的识别结果相近且优于动态时间规划;当训练样本集较小时,支持向量机的识别率最高。在训练和识别效率方面,支持向量机具有明显的优势。

To further improve the effect of cough automatic identification , support vector machine is adopted as classification mod-el for cough recognition .The process of sample collection , MFCC feature extraction and support vector machine cough recognition is introduced in detail , and the results are compared with hidden Markov model and dynamic time warping .Experiment results show that, with a big training sample set , recognition rates of support vector machine are similar with hidden Markov model and higher than dynamic time warping , while with a small training sample set , support vector machine achieves the best result .In terms of efficiency of the algorithm , support vector machine significantly outperforms the other two classification models in both training and recognition time .