Font Size: a A A

Research Of Speech Enhancement Based On Empirical Mode Decomposition

Posted on:2012-12-27Degree:MasterType:Thesis
Country:ChinaCandidate:M L SunFull Text:PDF
GTID:2218330368982548Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
The research of speech enhancement technology is more and more popular, with the rapid development of communication technology. Speech enhancement can filter the noisy speech signal to improve the quality of it, and make the speech more clearly and intelligibility so as to be accepted easily. Speech enhancement is widely application in speech recognition and coding, speech community and other fields. But in the process of speech communication, the speech signals are often corrupted unavoidable by various noises, which make the result in a noisy speech signal instead of the original purity speech signal. Therefore, the research on the technology of speech enhancement plays an important role in the practical life and plays a great value for society.There are varieties of speech enhancement algorithms. In this paper, the technology of speech enhancement method based on the empirical mode decomposition (EMD) is researched. The EMD popular and has a good performance in speech enhancement field, which is a new signal processing method. Since it is based on local chacteristic of a signal and adaptive, it is highly efficient and especially suitable for analyzing nonlinear and non-stationary signal. This method has an important theoretical value and wide application prospects. Currently it is widely used in speech enhancement field.This paper elaborates the basic concepts of speech enhancement and related theories, the basic principle and the steps of EMD.Summarized the major problems in EMD, included curve fitting, endpoint effects, mode mixing and step criterion. Then does depth research in speech enhancement based on EMD. For the problem of endpoint effect, study the improve algorithm extremum field mean mode decomposition (EMMD), and proposed the speech enhancement method based on EMMD. Experiment results show that the EMMD method can eliminate the DC-component implied in local data, restrain the endpoint in EMD effectively and is capable to improve the SNR.For the existence musical noise in EMMD method, make some depth research in speech enhancement based on combine EMMD with MMSE. Through theoretical analysis and experimental show that in low SNR conditions, the algorithm is performance good quality, which is not only can improve the SNR, but also effectively improve the clarity and intelligibility of the speech signal. Through make comparison with the method based on spectral subtraction, EMD, EMMD, we can come to the conclusion that the algorithm proposed in this paper is superior to other method in the signal processing field.
Keywords/Search Tags:Speech Enhancement, Empirical Mode Decomposition, Extremum Field Mean Mode Decomposition, Minimum Mean Square Error
PDF Full Text Request
Related items