A novel voice activity detection method using energy statistical complexity

In this paper, the nonlinear dynamic characteristics of the statistical complexity were applied to the voice activity detection (VAD). By combining it with the energy feature, we present a new VAD method that is energy statistics complexity (ESC) algorithm, using fuzzy c-Means clustering algorithm and Bayesian information criterion algorithm to estimate the thresholds of the ESC characteristic, and using dual threshold method for VAD. Experiments on the TIMIT continuous speech database show that at low SNR environments, ESC method is superior to the energy spectrum entropy (ESE) method. Especially in the vehicle noise and vehicle interior noise environments, ESC method shows better detection performance.