Feature selection in speech and speaker recognition by