Study of Improved Feature Technology for Abnormal Sound Recognition Based on Empirical Mode Decomposition

Abstract:

There has been increasing attention to the abnormal sound in our daily life, its identification technology has become a current research focus of non-speech signals. Aiming at abnormal sound detection and recognition in a large parking lot, this paper put forward a new approach to the feature extraction process of advanced Mel frequency cepstral coefficients (MFCC) based on empirical mode decomposition (EMD) for abnormal voice recognition, and Gaussian mixture model is used to design the classifier, implementing the classification for abnormal sound recognition of peoples scream, glass broken, car alarm and crash sound in the parking lot. The results demonstrated that the proposed of improved MFCC feature extraction algorithm can get more dynamic information ,hence have a higher recognition rate for abnormal sound, and the new characteristic feature has better robustness and accuracy under different signal-to-noise ratio.

• Periodical: Applied Mechanics and Materials (Volume 391)
• Pages: 576-579
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.391.576
• Online since: September 2013
• Authors: Hui Wang, Liang Wang, Yu Li Jun, Shao Ying Liu
• Keywords: Abnormal Sound Recognition, Empirical Mode Decomposition (EMD), Gaussian Mixture Model, Signal-to-Noise Ratio (SNR)

References

[1] Luan Shaowen, Gong Weiguo . Feature Extraction of Typical Abnormal Sounds in Public Places[J] . Computer Engineering. April 2010, 36(7): 208-210.

[2] Li Lingli, Chen Xiaoming. Application of class-dependent feature extraction and classification on environmental sounds[J]. Computer Engineering and Applications, 2012, 48(22): 156-160.

[3] Huang N E, Shen Z, Steven R Long. The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis[J]. Proc R Soc Lond A, 1998, 454: 903-995.

DOI: 10.1098/rspa.1998.0193

[4] Huang N E. Review of empirical mode decomposition[J]. Proceedings of SPIE-The international Society for Optical Engineering, 2001, 439(1): 71-80.

[5] Wang ZaiHuan, Tang YunJian, Han Peng. Method of detecting about unlawful cutting of forest based on audio recognition . Computer Engineering and Applications.

[6] F. Beritelli,S. Casale,A. Russo,S. Serrano,D. Ettorre, Speech emotion recognition using MFCCs extracted from a mobile terminal based on ETSI front end, in: 8th International Conference on Signal Processing, Beijing, (2006).

DOI: 10.1109/icosp.2006.345670

[7] E.K. Lenzi R.S. Mendes L.R. da Silva, Statistical mechanics based on Renyi entropy, Physica A: Statistical Mechanics and Its Applications 280 (3–4) (2000) 337–345.

DOI: 10.1016/s0378-4371(00)00007-8

[8] H. Mohammadi,R. Saeidi, Speaker identification performance enhancement using Gaussian mixture model with GMM classification post-processor, IEEE International Conference on Signal Processing and Communication (2007) 24–27.

DOI: 10.1109/icspc.2007.4728366