Chord Recognition With Beat Detection

Computers promote the development of modern electronic music technology. Music recognition,classification and feature extraction,such as a series of questions,which base on computer and electronic technology, have been increasingly concerned about by many researchers. Music recognition is a rising interdisciplinary research field,which involves physics,signals processing,human-computer interaction,music theory and music psychics. This dissertation combines the technique of multimedia, signal processing and pattern recognitions with music theory to make the computer to imitate the course of music cognitive and analysis of human. Beat is a fundamental unit of the temporal structure of music and automatic beat tracking forms the basis of a number of applications. paper,we present all automatic beat tracking system that processes music recordings and determine the temporal beat location of the music with kinds of genres.First,the formatted audio signals of polyphonic music are input into a front-end analysis that extracts CQT features,and features are transformed into onset sets.Then a tempo detection block detects the tempo by periodicity estimation,incorporating primitive musicological knowledge of tempo.At last,assumed that tempo is constant throughout a piece,the temporal estimation of beat locations is obtained by using dynamic programming to search all possible sets of beat instants and find the best global sequence of beat times.Secondly,the paper studied the Pitch Class Profile features and the calculation method which is often the choice of the feature in automatic chord recognition in western music,calculation formula is list out. We construct the music chord recognition system based on HMM,define 36-state for the HMM,each state represents a single chord. The observation distribution is modeled by a single multivariate Gaussian in 12 dimensions defined by its mean vector and covariance matrix. We get the supervised HMM by virtue of the label files made by Chris Harte. In the course of recognition,viterbi algorithm is applied to the model to find the optimal path,i.e,chord sequence,in a maximum likelihood sense given an input signal. The experimental recognize the chord type of three different music in frame-level,final average recognition accuracy is 75.56%. The paper discussed the experimental process and results in detail at last.