Algorithm Study Of Speech Enhancement Based On Self-Adaptive Filter And Implementation By DSP

The primary purpose of speech enhancement is to extract more puree speech from speech signal with noise. In this paper, a system of speech enhancement using DSP is discussed based on self-adaptive algorithm. On the basis of presenting an overview of a number of speech enhancement algorithm in common use, the self-adaptive system for speech enhancement is described in detail based LMS algorithm, which can get all-sided information of noise and better noise reduction effect with one reference noise as accessorial input. Because the LMS algorithm is simple, small in calculation and easy to realize in real time signal processing, LMS algorithm is widely used in the technology of speech enhancement.On the basis of analyzing and studying LMS algorithm, a simulation of the self-adaptive speech enhancement was made with Matlab and C++ language. For realizing the application of the speech enhancement with LMS algorithm, a algorithm platform was set up with DSP chip TMS320VC5402 and peripheral circuits in the system, which include speech input, AID conversion, self_adaptive treatment, D/A conversion and speech output. C Program is given and down loaded into DSP chip for the purpose of speech signal input, treatment and enhancement. The result shows that the algorithm can effectively suppress noise in speech , improve SNR and reduce distortion.It is one of the extensively adopted technological means using DSP processor toset up a speech processing platform. At present, to look for a suitable speechprocessing algorithm is also a hot problem in the speech processing field. We did a little study in this respect and made preliminary experience. All this will be the certain foundation for the further study in the future.