Automatic rhythm analysis is a development tool that have more and more popular in many applications, especially in computer music, accompaniment systems, musicology and automatic music transcription. The main goal of this thesis is to implement a real-time analysis tool that can perceive the rhythm. For example, we can apply real-time computer graphics synchronized with music and displays virtual dancers whose motions and positions change in time to beats. Beat detection plays an important role on this application. This research focuses on the related studies and implementation of beat detection on rhythm analysis. This thesis will introduce the background knowledge for human rhythm perception and some related speech signal processing techniques that will be applied to emulate rhythm perception. To decrease the operating time, our algorithm has be implemented to need only the integer operation. We also demonstrate the experimental results obtained in detecting the musical beats from audio signals sampled form compact discs of technos and pop musics. Finally, we will address our future works and give some directions on future researches.