Based On Chaotic Systems Analysis Of Nonlinear Speech Coding

Though the speech production is nonlinear and chaotic, the traditional speech processing methods are mainly based on the deterministic linear system theory. With the development of chaotic dynamics, people's visions are greatly broadened, and it provides a new way to explore the complex natural phenomena. Chaos theory has greatly affected the research on natural sciences and social science. In recent years, many researchers have concerned the application of nonlinear models in speech processing. As an application of chaos theory, paper studied the nonlinear characteristics of speech, established a nonlinear predictor based on the analysis of chaotic system, and also established a speech coding system using the nonlinear predictor.First, papers introduce the knowledge of chaos and the characterization of chaos. On this basis, paper studied the chaotic characterization of speech signal, and studied the Chinese phoneme using fractal dimensions, Lyapunov exponent, the least embedding dimension, as well as the reconstructed phase space.Then, based on the nonlinear analysis of speech, papers established a nonlinear predictor using RBF networks, and studied its structure and training algorithm. Compared with the linear predictor, the nonlinear one has better performance.At last, papers provide a nonlinear predictive coding system using the nonlinear predictor instead the linear one. Simulation results show that the system can improve the coding performance.