Combining the prediction scores of recommendation items based on content-based recommendation algorithm (CB) for weighted mixing, using stochastic gradient descent to solve the linear regression problem in the weighted mixed recommendation model to obtain the optimal weighted parameters, and using web crawling technology to obtain massive music data and front-end and back-end technology to design the interface of the music system. The multi strategy fusion recommendation algorithm is applied to the music recommendation system. Tests have shown that the accuracy and recall of the recommendation algorithm have reached 87.63% and 48.65%, respectively. Based on historical usage data, the average accurate recommendation rate per user is calculated to be 86.72%, successfully confirming the feasibility of implementing the multi strategy fusion recommendation algorithm on music recommendation systems.