Research of a Non-Specific Person Noise-Robust Speech Recognition System

To solve the problem that the performance of speech recognition systems declines in the noisy environment, this paper used the linear predictive Mel frequency cepstrum coefficients according with human hearings characteristic as speech feature parameters, adopted two recognition machines, the support vector machine and the wavelet neural network, realized respectively a speech recognition system of non-specific person and isolated words with visual C++ programming, got the recognition correct rates in different SNRs and in different words, and compared their recognition results with those of based on traditional hidden Markov models. Experiments indicate that the recognition correct rates based on the support vector machine and the wavelet neural network are all higher than based on traditional hidden Markov models, and also have better robustness.