Speech Recognition in Aphasiac Rehabilitation Based on Wavelet

The abstract of real voice of the aphasiacs is of great importance in the rehabilitation system. In this paper a new method of using wavelet for rehabilitation system was raised, it also improves the wavelet decomposition by refining its high frequency part in order to improve the SNR of voices. Simulation was given by using the combination of advanced wavelet decomposition and soft thresholding. The experiment results showed that this new method is efficient in enhancing the SNR and reserving the energy of effective signals. It supplements the deficiency of soft thresholding and aquires good results without noises. It has practical significance for auditory rehabilitation system for aphasiac patients.