The Speech Analyzing And Training System Base On Thewavelet Transform

China is one of the countries where the speech disease occupies a high proportion of the population. If those speech diseases have not received special speech training, they are likely to lose the natural ability of speech communication. If developing the special software by the multimedia of computer, we can train and cure speech disease patient. By using speech process technology speech signal changed to different images. The patient can judge the accuracy rating of pronounce by these images, and correct pronounce.In this thesis, it is firstly summarized that some basic knowledge of speech signal, and then it is introduced that the speech analysis methods in time and frequency domain. The time domain analysis is most simple and intuitionistic. The short-time energy, the short-time zero crossing rate and the short-time self-correlation are main analysis method in time domain. The frequency domain analysis can get many characters which can not get by the time domain analysis. We can get frequency information of speech through the analysis of the spectrogram and spectrum which get by using the short term Fourier transform.The spectrogram can express the relationship between time and frequency domain analysis, and it is main analysis method. We introduce that the spectrogram can get by using the wavelet transform. The spectrogram which get by using the wavelet transform has good frequency resolution in low frequency and good time resolution in high frequency, this more accord with the characters of speech. We analyzed a lot of speech, and the results show the wavelet transform has more advantage.At last, it is introduced that the realization of a speech analyzing and training system. The system is based on wavelet transform and other time and frequency domain analysis method. The control menu is developed in Visual C++ 6.0, and the analysis program is developed in Matlab 6.5. The analysis program is called in Visual C++, and the developing time is shortened. The system can control the playing, recording and pausing, and can display figure forms of the input speech both in time domain and frequency domain. The user can set up different parameter because of different needs, so they can get more exact result. Those inarticulate trainees can carry out speech training by comparing spectrogram as visual feedback.