Abstract

In this paper, an improved a priori SNR estimator is presented for speech enhancement in hands-free communication systems. The proposed a priori SNR estimation technique, based upon the Two-Step Noise Reduction (TSNR) algorithm, has flexibility and control over noise attenuation performance and speech distortion reduction. In addition, an efficient technique for estimating the Echo Spectrum Variance (ESV) is introduced. Simulation Results demonstrate significant improvment of the Acoustic Echo and Noise Suppression (AENS) systems in terms of Noise and Echo Attenuation performance and its ability to preserve the enhanced speech quality.