A Research Of The Speech Recognition Based On The SOM Model

Speech recognition is a complex nonlinear process. Up to now, most speech recognition methods are based on conventional linear system theory, such as Hidden Markov Model (HMM) are hard to have a breakthrough. Recently, with the development of nonlinear-system theories about artificial neural networks (ANN), the recognition method based on ANN became the focus of the research.The research in this paper is oriented on the theory and application in the speech recognition of the Self-Organizing feature Map (SOM), and the related algorithms and model are developed. The design and exploitation of software for experiments are also completed based on MATLAB7.0.This paper discussed the technology of linear prediction code and the bandpass filters analysis method, then reduced the LPC, LPCC and the MFCC parameters. The design principle of SOM and the effects of different feature parameters to speech recognition results are analyzed and discussed. Through the further training method, the recognition capability of the SOM is improved much more. Experiment result show better recognition performance and particular application advantages are achieved by the method for speech recognition based on SOM. Especially for the Isolated Word Recognition (IWR), the recognition accuracy is over than 95%.