The article researches on the application of adaptive filtering theory on the noise processing of speech signal based on the characteristics analysis of speech signal.The article combines the digital subsidence filtering and adaptive filtering and builds up different reference models to accomplish a adaptive process according to signals of different frequences. We bring forward the method through compare the two projects adaptive spectrum enhancement and wavelet analysis which combines the wavelet analysis and the adaptive theory and have the information of several frequence bands being the reference input signals to reach the adaptive processing and restructure of speech signal by making use of merits of the two theories. The transcribed voice signals are transformed into discrete digital signals with the matlab and recorder of windows to collect and deal with speech signals. During the procedure a lot of experiments and emulation are made.
↧