Abstract: A method based on broadband and narrowband spectrogram fusion with zoning projection of specific two words Chinese lexical recognition was presented.In the process of image feature extraction,the image processing technique is applied to the speech recognition field.Firstly,equal width zoning line projection and binary width zoning line projection are carried out to the narrowband spectrogram,and they are set respectively as the narrowband spectrogram of the first characteristic set and the second characteristic set.Meanwhile,equal width zoning line projection is carried out again to the narrowband spectrogram after Fourier transform,treating it as the third feature set.Then,equal width column projection is carried out to the broadband spectrogram,regarding it as the fourth feature set.The above three feature sets are used as feature vectors to support vector machine(SVM) as a classifier for the overall recognition of specific two words Chinese vocabulary.1000 voice samples are used in simulation experiment.The results show that the correct recognition rate of the two words Chinese word recognition by the first three feature sets is 92.4％.The correct recognition rate of two words vocabulary recognition using fourth feature sets is 80％.The correct recognition rate of the two words Chinese word recognition by using the feature value fusion of the above four features can reach 95.4％.This method of feature fusion provides a new way of thinking of Chinese vocabulary overall recognition.

Key words: Speech recognition,Spectrogram,Feature fusion,Line projection,Column projection,Support vector machine(SVM)