Method and apparatus for speaker recognition through testing of oral information by means of forced decoding