Voice Activity Detection Based on Auto-Correlation Function Using Wavelet Transform and Teager Energy Operator

Computational Linguistics and Chinese Language Processing Vol. 11, No. 1, March 2006, pp. 87-100 87 © The Association for Computational Linguistics and Chinese Language Processing [Received May 13, 2005; Revised Oct. 8, 2005; Accepted Oct. 17, 2005] Voice Activity Detection Based on Auto-Correlation Function Using Wavelet Transform and Teager Energy Operator Bing-Fei Wu∗ and Kun-Ching Wang+ Abstract In this paper, a new robust wavelet-based voice activity detection (VAD) algorithm derived from the discrete wavelet transform (DWT) and Teager energy operation (TEO) processing is presented. We decompose the speech signal into four subbands by using the DWT. By means of the multi-resolution analysis property of the DWT, the voiced, unvoiced, and transient components of speech can be distinctly discriminated. In order to develop a robust feature parameter called the speech activity envelope (SAE), the TEO is then applied to the DWT coefficients of each subband. The periodicity of speech signal is further exploited by using the subband signal auto-correlation function (SSACF) for. Experimental results show that the proposed SAE feature parameter can extract the speech activity under poor SNR conditions and that it is also insensitive to variable-level of noise. Keywords: Voice Activity Detection, Auto-Correlation, Wavelet, Teager Energy 1. Introduction Voice activity detection (VAD) refers to the ability to distinguish speech from noise and is an integral part of a variety of speech communication systems, such as speech coding, speech recognition, hand-free telephony, and echo cancellation. In the GSM-based communication system, a VAD scheme is used to lengthen the battery power through discontinuous transmission when speech-pause is detected [Freeman et al. 1989]. Moreover, a VAD algorithm can be used under a variable bit rate of the speech coding system in order to control the average bit rate and the overall quality of speech coding [Kondoz et al. 1994]. Perviously, ∗ Department of Electrical and Control Engineering, National Chiao-Tung University, HsinChu, Taiwan E-mail: bwu@cssp.cn.nctu.edu.tw + Information & Communications Research Laboratories, Industrial Technology Research Institute, HsinChu, Taiwan E-mail: kunching@itri.org.tw 88 Bing-Fei Wu and Kun-Ching Wang Sohn et al. [Sohn et al. 1998] presented a VAD algorithm that adopts a novel noise spectrum adaptation by applying soft decision techniques. The decision rule is drawn from the generalized likelihood ratio test by assuming that the noise statistics are known a priori. Cho et al. [Cho et al. 2001] presented an improved version of the algorithm designed by Sohn. Specifically, Cho presented a smoothed likelihood ratio test to reduce the detection errors. Furthermore, Beritelli et al. [Beritelli et al. 1998] developed a fuzzy VAD using a pattern matching block consisting of a set of six fuzzy rules. Additionally, Nemer et al. [Nemer et al. 2001] designed a robust algorithm based on higher order statistics (HOS) in the residual domain of the linear prediction coding coefficients (LPC). Meanwhile, the International Telecommunication Union-Telecommunications Sector (ITU-T) designed G. 729B VAD [Benyassine et al. 1997], which consists of a set of metrics, including line spectral frequencies (LSF), low band energy, zero-crossing rate (ZCR), and full-band energy. However, the common feature …

Content truncated to 500 words