| Speech is one of the best natural and convenient intercommunication manners among humankind. It is also one of the ideal man-machine conversation interfaces. Nowadays, Speech processing technologies have been broadly used in many applied fields. In this dissertation, the main research focus is on the study of speech enhancement and separation, which is one of the key technologies when we try to put the speech processing into reality. Speech enhancement and separation deals with the problem of how to enhance or separate the useful speech signals from the mixed inputs under various noisy background enviro- nments, especially when the interfering signals have speech-like statistical property.Firstly, The background noise has destroyed the already existing acoustics characteristic of speech signal and model parameter. In requests to enhance the speech communications quality demand, along with the digital signal processing technology rapid development, the chip price drop, enable the research to have the vital significance take the new hardware new algorithm as the foundation new speech enhancement method.Secondly,This paper mainly discusses speech enhancement base on neural network. First introduced speech single as well as the neural network elementary theory, then proposed based on the BP neural network speech enhancement system modeling method. Synthesizes the speech feature extraction, induced the speech cepstrum and the Noise cepstrum estimate value uses in the neural network training and the study and finally eliminates the noise, the experiment proved this method had to surpass traditional the speech enhancement algorithm. The simulation result show, this speech enhancement system design method, not only realizes simply, but also saves the running time, The speech enhancement effect is good.Thirdly,This article discusses a new method of speech enhance based on neural network. MFCC coefficient of the speech signal can be picked up under noise and... |
PDF Full Text Request