Nonlinear Parameter based Endpoint Detection Algorithm for Isolated Word Recognition System with Neural Network
장한, 정길도 · 대한전자공학회 학술대회 · 2010
In this paper, with a view to the chaos characteristic of speech signal, which is ignored in conventional research, we explore the extraction of an important nonlinear parameter, i.e. time-dependent Lyapunov Exponents (LEs) to realize the endpoint detection of speech signal. The simulation results show higher detection accuracy for each Isolated Word, also the contribution to facilitate the subsequent network training to guarantee high recognition rate.