In this paper proposed speaker gender identification algorithm, 
based on the binarization of acoustic features. 
The error rate of identification for male is 0.57%, for female - 1.10%.