Humming computer music searching method based on longest matching subsequence algorithm

Abstract

The invention discloses a humming computer music searching method based on longest matching subsequence algorithm, which comprises the following steps of (1) fundamental tone frequency extraction, (2) music characteristic database construction, (3) characteristic expression achieving and (4) searching matching. The method has the advantages of improving integral speed of similarity calculation, improving searching efficiency of a searching engine, constructing an accurate music searching platform for karaoke, a network engine based on content search and a multifunctional intelligent mobile terminal platform and being capable of being widely used in fields of relative plugs of the network searching engine and the like. The method for music characteristic extraction, music characteristic expression and similarity accurate calculation can conduct accurate calculation on a humming searching system, enables music search to be accurate, slight and happy, and has strong practical value and reality meaning.

Description

technical field [0001] The invention relates to a humming computer music retrieval method based on the longest matching subsequence algorithm, and belongs to the technical field of computer application based on music information content retrieval. Background technique [0002] In recent years, with the development of the Internet, audio data has grown exponentially. Traditional text-based retrieval methods can no longer meet the retrieval needs of massive multimedia data, so content-based Music Information Retrieval (MIR) technology has become one of the hot technologies in the fields of signal processing, pattern recognition, and data mining. . The research of content-based multimedia information retrieval technology mainly focuses on image and video. At present, there are not many technologies applied to audio retrieval at home and abroad. As users' interest in web classification and retrieval increases, it is crucial to establish an audio web data retrieval mechanism. ...

AI Technical Summary

Problems solved by technology

[0005] 1. The pitch contour expression method cannot quantify the pitch change, and it is easy to cause the feature expression and the semantic content of real music to be blurred. With the expansion of song samples, it is easy to have the same pitch contour but the actual melody is very different.

[0006] 2. The approximate MIDI note expression method approximates the natural pitch of the user's humming to the integer value of the discrete MIDI note, which will cause the problem of inaccurate melody expression

In a reasonable feature expression method, it should be considered that the two melodies have the same melody feature; based on this, it shows that the approximate expression method of MIDI notes is not appropriate and comprehensive.

[0007] 3. Although the absolute pitch expression method solves the problem of expression errors caused by approximation, the pitch shiftiness (Pitch Shiftiness) produced when combined with the string comparison class and some dynamic programming related algorithms will cause serious problems. Matching error, so this feature expression method is not suitable for general similarity calculation mechanism

But at the same time, it also has unavoidable shortcomings: Hidden Markov model needs to establish a corresponding training model for each record in the music feature database. As the capacity of the feature database increases, the workload of training will be very large. Markov models are less practical

Images