The invention discloses a music recommendation method based on group perspective, and particularly relates to the technical field of music recommendation. The method specifically comprises the following steps: obtaining static information and dynamic information of a user portrait, sending a table and a questionnaire to a user to obtain the static information of the user portrait, carrying out statistics and processing on data generated by user behaviors regularly, obtaining the dynamic information of the user portrait, and updating a dynamic information label of the user portrait periodically; constructing a user portrait after the dynamic information and the static information of the user are processed, abstracting information characteristics of a single user through the user portrait, tagging the user, and describing each user with a plurality of labels at different angles. According to the method, the real-time user dynamic portrait information can be acquired, and the music pushing is performed according to the real-time user dynamic portrait information, so that the pushed music content style can be changed according to the change of the user dynamic portrait information, the music pushing precision is improved, and the use experience of the user is improved.