Song recommendation method and system

Abstract

The invention relates to a song recommendation method and system. The method comprises the steps of obtaining a corresponding user song-listening preference model according to song-listening behaviorinformation of a user and song-listening source information; based on a preset rule, obtaining a corresponding song audio feature correlation model; according to the user song-listening preference model, selecting a song preferred by the user, obtaining corresponding songs with similar audios according to the song audio feature correlation model for the selected current song, and scoring the songswith the similar audios to obtain corresponding scores; and sorting the scores of the songs with the similar audios, and preferentially recommending the similar songs with the high scores. The song audio feature correlation model enriches a similar song database capable of performing song recommendation; the problem of few songs in an existing similar song database is solved; and furthermore, based on the similarity of the songs in audio features including beats, tone, timbre and the like, song recommendation is performed, so that the song recommendation accuracy is improved.

Description

technical field [0001] The present invention relates to the field of audio technology, in particular, the present invention relates to a song recommendation method and system. Background technique [0002] At present, there are often two common song recommendation methods, one is a song recommendation method based on item-based collaborative filtering, and the other is a song recommendation method based on song content. [0003] First of all, the principle of item-based collaborative filtering song recommendation method is that it is used to identify songs that a specific customer may be interested in at present for song recommendation. The conclusion of recommending songs for a specific customer is based on the analysis of which songs other similar customers are interested in. [0004] The song recommendation method based on item-based collaborative filtering analyzes the user's interest in listening to songs, finds users with similar interests to a specific user in the us...

AI Technical Summary

Problems solved by technology

[0008] The embodiment of the present invention is to provide a song recommendation method and system. Through the song audio feature correlation model, the similar song database that can be recommended for songs is enriched, and the problem of the small number of songs in the current similar song database is solved. Further, Based on the similarity of songs in beat, pitch, timbre and other audio characteristics, song recommendation improves the accuracy of song recommendation

Images