The Research On The Key Technology Of Web-based Query By Humming

With the development of internet, more and more people hope to search the things interesting them, such as music and pictures, which raises the requirement for multi-media search.Audio retrieval is a new focus after image retrieval in the field of Content-Based Retrieval. As a branch of Content-Based Audio Retrieval, Content-Based Music Retrieval Query by Humming offers a totally new retrieval method by searching the desired song based on the user's humming segments in large data base quickly. It will improve the user's interactive experience by applying this new music retrieval method on the internet.This paper makes the following research, focusing on the key technology of Web-based Query by Humming.(1) Studies the relevant theories of sound signal , laying the foundation for future treatment of music signals and system realization.(2) Studies the fundamental characteristics of music. Realizes a pitch period measurement solution based on the rhyme of the music signal and giving priority to the expression of music rhyme and pitch measurement. Offers the extraction arithmetic based on melody contour through studying the extraction arithmetic of music rhyme features.(3) Adopts similarity-matching method based on dynamic programming to retrieve music by describing and analyzing the Content-based Music Retrieval arithmetic and music similarity calculation with the combination of the expression of the music features and the error analysis of humming input.(4) Offers the framework of Web-based Query by Humming system and designs the database of the music character. Realizes the Web-based Query by Humming system, analyse the experiment result which validates the the feasibility and effectiveness of the system.