关键词: Mel频率倒谱系数, 支持向量机, 交叉验证, 环境音分类, 特征提取

Abstract: In general, recognition methods applied for music and speech data are not appropriate for the environmental sounds. In this paper, we propose a MFCC (Mel frequency cepstrum coefficients)-SVM (support vector machine)-based approach that exploits feature representation and learner optimization to achieve the classification of 10 different environmental sounds signals in the office. Environmental sounds events are obtained by using the IEEE AASP (Audio and Acoustic Signal Processing) Challenge Dataset. The proposed approach considers efficient representation of MFCC features by changing the number of Mel coefficients in analyzing as well as optimizing the SVM parameters. Experiment shows that, when the results of the proposed methods are chosen for MFFC feature and SVM classifier, the tests conducted through using 5-fold cross validation, the average classification accuracy can be up to 88.05%. The classification effect is significantly better than the default MFCC-SVM algorithm.

Key words: Mel frequency cepstrum coefficients (MFCC), support vector machine (SVM), cross validation, environmental sounds classification, feature extraction