Feature extraction approach towards assuring and improving the quality of speaker recognition performance