Speech recognition process is kind of applied Technique of digital sign process that is widely used for many applications, for example technology in the field of telecommunications which isn't only able to provide for serving for sending text data but also it can serve for sending data using sound. From technology development of signal process emerges new idea to make application program by creating new sofware to display sound signal characteristic based on frequency and highest magnitude. Clustering is part of pattern recognition science which is made for system that can perform into a groups. In this research, the researcher will distinguish between male voice or female voice. Mechanism process uses Collecting voice samples, then the feature extraction using FFT that produce two main features that is maximum value of frequency and maximun value of magnitude. After that, k-means alogarithm process is used for grouping voice of male clusteror female cluster. The result of this research uses 20 training data and 20 testing data that will produce 75% level of accuracy for training data and 100% for testing data.