Content-Based Music Retrieval Algorithm and Simulation Analysis

To quickly and accurately retrieve the required music songs from the mass music database, this paper introduces the framework of content-based audio retrieval system and its related characteristics and difficulties. Then it describes the process of human voice humming feature processing on the voice platform within the framework of the system. On the simulation music platform, the retrieval algorithm based on the combination of notes and fundamental frequency is studied and analyzed for songs with background music, combined with the relevant characteristics of audio signals. The accuracy of humming retrieval can be improved by using more accurate iterative alignment algorithm of stress shift. The experiments show that the average retrieval time of humming retrieval system is significantly reduced by using the retrieval strategy.