An efficient VAD based on a hang-over scheme and a likelihood ratio test

Abstract

The emerging applications of wireless speech communication are demanding increasing levels of performance in noise adverse environments together with the design of high response rate speech processing systems. This is a serious obstacle to meet the demands of modern applications and therefore these systems often needs a noise reduction algorithm working in combination with a precise voice activity detector (VAD). This paper presents a new voice activity detector (VAD) for improving speech detection robustness in noisy environments and the performance of speech recognition systems. The algorithm defines an optimum likelihood ratio test (LRT) involving Multiple and correlated Observations (MO) and assuming a jointly Gaussian probability density function (jGpdf). An analysis of the methodology for N = {2,3} shows the robustness of the proposed approach by means of a clear reduction of the classification error as the number of observations is increased. The algorithm is also compared to different VAD methods including the G.729, AMR and AFE standards, as well as recently reported algorithms showing a sustained advantage in speech/non-speech detection accuracy and speech recognition performance. © Springer-Verlag Berlin Heidelberg 2007.