Investigation of the speaker identification method based on clustered pseudostationary segments of voiced sounds