Artificial Neural Network was used to clariffy recorded human voices based on the gender of the speaker. Due to the presence of surrounding noises and inaccuracy during the record of the voices, preprocessing of the data is essential to minize noise and exclude the irrelevent parts. There are various ways to preprocess data but in this project, only three are chosen. The three ways used are minizing the non-speech content, eliminate the DC content and normalization of the data. Fast Fourier Transform was then used to prepare the data in a form suitable for the training of ANN. The trained system is then used to determine of the gender of voices that are fed into it.