文章基本信息

标题：Noise Reduction Using Mel-Scale Spectral Subtraction with Perceptually Defined Subtraction Parameters - A New Scheme
本地全文：下载
作者：Ch. V. Rama Rao ; M. B. Rama Murthy ; K. Srinivasa Rao 等
期刊名称：Signal & Image Processing : An International Journal (SIPIJ)
印刷版ISSN：2229-3922
电子版ISSN：0976-710X
出版年度：2011
卷号：2
期号：1
页码：135
出版社：Academy & Industry Research Collaboration Center (AIRCC)
摘要：The noise signal does not affect uniformly the speech signal over the whole spectrum isn the case ofcolored noise. In order to deal with speech improvement in such situations a new spectral subtractionalgorithm is proposed for reducing colored noise from noise corrupted speech. The spectrum is dividedinto frequency sub-bands based on a nonlinear multiband bark scale. For each sub-band, the noisecorrupted speech power in past and present time frames is compared to statistics of the noise power toimprove the determination of the presence or absence of speech. During the subtraction process, a largerproportion of noise is removed from sub-bands that do not contain speech. For sub-bands that containspeech, a function is developed which allows for the removal of less noise during relatively low amplitudespeech and more noise during relatively high amplitude speech .Further the performance of the spectralsubtraction is improved by formulating process without neglecting the cross correlation between the speechsignal and background noise. Residual noise can be masked by exploiting the masking properties of thehuman auditory system. In the proposed method subtraction parameters are adaptively adjusted usingnoise masking threshold. A psychoacoustically motivated weighting filter was included to eliminateresidual musical noise. Experimental results show that the algorithm removes more colored noise withoutremoving the relatively low amplitude speech at the beginning and ending of words.
关键词：Speech enhancement; Spectral subtraction; Cross correlation; Sub-bands