We successfully wrote a program which inputs and analyzes spectrographic points of a human voice in a Fast

        Fourier Transform equation on the supercomputer (see Appendix B). Although the process of getting the points from the

        WAV file to the supercomputer was rather tedious and manual, the program did its job. But using the method of average

        summation to compare voice files proved to be very ineffective due to our incorrect assumption that using the FFT would

        calculate and identify a voice. Since the fluctuation in points was so great, the average summation of one person’s voice

        was significantly different for every different sample of that person’s voice. We are currently working at a method of

        comparing points individually instead of on an average basis. We feel that if this can be accomplished, we will obtain an

        identification accuracy exceeding 50%.