Research On Recognition Algorithm Of Piano Music And The APP Implementation

Automatic recognition of music,as a new mixed discipline,has very important application in the field of music retrieval and automatic music composition.At present,the research of music recognition mainly focuses on the recognition of single note,and there are some limitations in recognition accuracy and anti-noise performance.Based on the deep understanding of the music system,some key techniques of piano music recognition are discussed,such as detection of start and end of a note,accurate extraction of frame-based frequency,precise calculation of fundamental frequency and so on,and then a solutions for recognition of piano music with continuous notes is proposed.The main contribution of this paper is as follows:Firstly,paragraphs of music and noise are segmented based on the single-threshold energy difference method,then start and end of a note in the music are detected based on short-term energy difference method,in which characteristics of piano music are used to identify the energy jump point,and thus effectively improve the detecting accuracy in the start and end of the notes and avoid the situation of missed judgment and wrong judgment.Secondly,based on the theories of autocorrelation,cepstrum and short-range amplitude difference,an improved method to extract the base frequency of music samples is carried out,which can highlight the peak characteristic of the periodicity of frame samples,avoid the impact of half-frequency and multiplier and effectively improve the accuracy of the fundamental frequency extraction.Thirdly,on the basis of analyzing and comparing the waveform undulation characteristics of the music signal and the data processing method of the note frame sample,the method of calculating the fundamental frequency is improved,which achieves higher accuracy and fault tolerance by assigning a higher weight to a note's middle frame.Finally,a music recognition system based on Android mobile terminal + server is implemented,which is used to test the above-mentioned algorithms.The feasibility and efficiency is verified.