In this paper we present a prototype to discover the unsupervised repeating temporary perception in a time series. The purpose of this work is to control the case of random variable and to find out the measurements caused by the phenomena of simultaneous synchronization. The proposed model has used the non-parametric Bayesian technique to trace the motifs and their occurrences in the data documents. We introduce the Multivariate Rhythm Sequence Technique (MRST) method to find the rebound and repeated motifs and their instance in every document automatically and simultaneously. This model is used in wide range of applications and concentrates on datasets from different modalities.The video footages from non-dynamic cameras and data location bounded to the motif-mining server. The high semantic internal representation of the method gives advantage in operation such as event counting or analyse the scène. We used the sample images and videos from New York City traffic data for experiments with and the results shows better performance than the existing motif mixtures analysis in the time series.