Bandwidth Expansion Method Of Speech Based On Wavelet Transform Modulus Maxima

With the development of computer technology, in order to obtain a better voice quality under the current communication network condition, the technical workers proposed an artificial bandwidth extension method. The characteristic of this technology is:use narrow-band prior voice messages to recreate the wideband speech information. In this paper, a method of speech bandwidth expansion based on wavelet transform modulus maxima is proposed. By taking using of the similarity of the modulus maxima between narrowband wavelet analysis signals and wideband wavelet analysis signals, a mapping structure can be determined to perform the required bandwidth expansion given only the bandlimited speech signal version. Since the proposed method runs in the time domain, it provides a flexibility of frame selection whose length is variable that facilitates small time delay and potentially data-dependent speech segment processing to further improve the speech quality[421.Evaluations based on both objective and subjective measures show that the proposed bandwidth expansion approach results in high-quality synthesized wideband speech with little perceivable distortion from the original wideband speech signals. The major work is as follows:(1) The generation model of speech signal and pre-processing techniques are briefly introduced.(2) Some speech analytical methods which include cepstrum, linear prediction, wavelet analysis, vector quantization, and neural network are introduced.(3) A wavelet modulus maxima and neural network based speech bandwidth h extension method is proposed.(4) The wavelet modulus maxima and neural network based speech bandwidth extension method is implemented, and some simulations are presented. Simulation results reveal the validity of the proposed method.