Recommendation System has been widely used to personalize user experience on their own musical journey. With the advent of digital content distribution and cloud systems, we can capture user preference on an unprecedented scale. This report focuses on the two most ubiquitous types of approaches: Collaborative filtering and Content-based, to provide song recommendations for users based on past listening history and song characteristics. The collaborative-filtering hypotheses users with similar preferences now will hold similar opinions in the future. We use the Alternating Least Square (ALS) algorithm to recommend 10 songs for the top 3 active users by learning the user’s listening habits. The content-based approach is implemented through K-Means clustering techniques that identify the resemblance among songs. We utilized both the structure and unstructured features to find similarity. We retrieved the users’ top 3 songs and searched for the 4 most similar songs within their same clusters and use Manhattan Distance as a measurement.