Modeling Consistency in a Speaker Independent Continuous Speech Recognition System