期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2016
卷号：87
期号：2
出版社：Journal of Theoretical and Applied
摘要：In this paper, we present Gabor features extraction based on front-end processing using histogram equalization for noisy speech recognition. The proposed features named as Histogram Equalization of Gabor Bark Spectrum features, HeqGBS features are extracted using 2-D Gabor processing followed by a histogram equalization step from spectro-temporal representation of Bark spectrum of speech signal. The histogram equalization is used as front-end processing in order to reduce and eliminate the undesired information from the used spectrum representation. The proposed HeqGBS features are evaluated on recognition task of noisy isolated speech words using HMM. The obtained recognition rates confirm that the HeqGBS features yield interesting results compared to those of the Gabor features which are obtained from log Mel-spectrogram.
