Research And Implementation Of Query By Humming System In Music Player

Accompanied the rapid development of multimedia technology, coupled with the rapid spread of the network, large amount of music dates’inquiry and location as a problem has become the urgent need to address by people. Query-By-Humming is the study about Accurate and convenient positioning from a large of music’ data, it gradually becomes a hot research topic. Now we use the text-based music retrieval, this search can only be retrieved by entering the song name or artist name and other information. With the increase in the amount of music data and the strengthening of people to pursue the convenient concept of life, text-based music retrieval is very limited. If we do not know the song name or artist name, it is no way to research. However, content-based music retrieval will be able to solve this problems, it allows the users to retrieve the desired song by humming the song clips. The song or artist’ names and other text messages are not required. Require the users to do just humming out the parts of the song fragment he can remember. This new type of music’retrieval methods has revolutionary practical significance, and it also represents the unique characteristics of music. It has been a strong concern by Artificial Intelligence researchers and the general music lovers.This article firstly changes the MIDI format music to WAVE format music. Analyze the audio of WAV format and Extract information such as Audio melody. Extract and intercept Music melody of humming query fragments and in the standard database of music by the fundamental frequency extraction algorithm and endpoint detection method, and use N-Gram string matching algorithm to match the resulting music melody approximately to get approximate matching similarity. Then, Sort the similarity from high to low to use the DTW for detailed match. This article is in accordance with the basic idea of Query-By-Humming, Firstly study and analyz the theoretical knowledge of Query-By-Humming and related music audio processin technology. In the extraction of melodic features of humming-music and standar music, in accordance with their respective characteristics, use a different endpoi detection method to intercept the musical passages. Furthermore, Integrated us the extraction method of music melody feature of pitch characteristics an rhythm features. The method can be accurater and quicker to return to the retrieveabl information for the humming users. Finally, by studying and comparing the musi retrieval matching algorithm, use an improved hierarchical and adding the endpoir detection matching algorithm of Query-By-Humming which based on the Musi retrieval matching algorithm of N-Gram-string matching and sentence-length DTW.