The growth of digital music libraries shows the importance of having the best music to help users find customized content. This paper presents an integrated approach to music recommendation, as users receive content based on recommendations of songs that other users have already searched for and the content of song recommendations relevant to user’s taste. This article tried to use K-Means clustering technology to find similar users by content, used PLSA technology to separate active or most listened to songs by content. To accomplish this task, a set of many documents was collected, assuming that they could be used empirically. To evaluate the usefulness of recommendations generated using the music recommendation method, three parameters, including accuracy, F-1 score, and recall, were calculated. Results of the provided method showed good results by evaluating the parameters of the collected data (e.g., precision, recall, and F1 score). The evaluation can be done from the perspective of users’ real data, which will take some time due to the availability of data.