A Study on the Application of Textual Fingerprinting in Melodic Similarity Detection
Yang Li, Xien Cheng · 2024
Melody plagiarism is a serious problem, and realizing the similarity detection of music melody is extremely important for rectifying music plagiarism. In order to effectively recognize melody plagiarism, this paper proposes a melody similarity detection scheme based on text fingerprinting. The method converts the music melody into text, uses the N -gram algorithm to slice and hash encode the long melody segments, generates the text fingerprint by MinHash algorithm, and finally uses the LSH algorithm to perform fast similarity retrieval to realize the efficient matching between melodies. The experimental results show that the average accuracy of similarity recognition of this scheme reaches more than 90%, and it can effectively identify the music melodies with plagiarism.