Research-based Midi Instrument Control System And Notes Automatic Identification Method

The developments of the computer promote the development of modern electronic music. The technologies that base on the computer and electronic technology, music synthesis, music recognition and feature extraction technology are increasingly applied to the field of music. This article studies the subject proposed by a particular symphony orchestra. They hope to operate some large-scale instruments indirectly through a smaller piano keyboard, as well as to identify the music and automatically records the notes of music.Firstly the article discusses the design of a control system for the musical instruments implementation process. It introduces the way that the system exchanges the data via MIDI signal and the process that the PIC microcontroller handles the MIDI data flow. At last the problem of real-time and multi-threaded in microcontroller is discussed.Then the article studies the music recognition methods. First the basic characteristics of the music and basic music theory are introduced, as well as the traditional time domain and transform domain pitch identification method. Then two identification methods for a single note that base on two different principles are described in two separate chapters. The first method is based on pitch recognition in which the article discusses an improved algorithm: combination of autocorrelation and FFT-based recognition and how the short time and zero plot method can be used to achieve the segmentation of the music. The second method is based on the RBF neural network and the MFCC feature extraction, in which the article discusses the classification through neural network as well as the extraction of MFCC feature. Through the simulation and test, it is proved that note recognition method based on neural network is more reliable.Next this article will discuss how to apply the Dempster-Shafer theory of evidence on the fusion of the results of the two identify methods in order to make full use of the advantages of the two methods and get more the accurate results.Finally the article discusses the work that has been accomplished as well as the direction of further research.