Note Recognition Algorithm Research

Notes recognition system research began in the early 1970 s, the first recognition system appeared in 1975, 1966 system has been able to handle more complicated piano works. But these are still not mature. Compose music and, in the field of music creation, a lot of people come into contact with some of the speech recognition application software, combined with the intuitive feelings on the computer, they proposed an idea: if you can use the computer to automatically identify music from and can automatically complete the music creation, will make their work efficiency is greatly improved, and great inspiration, change the present for manuscript writing music creation inconvenience caused low efficiency even inspiration play a bad situation.This paper introduce the deep summary of the basic knowledge of music from the close to the distan, including a variety of sounds in the music and sound meaning. From the transition to the tonality, expounds the principle and characteristics of the piano notes twelve tone equal temperament. Then according to the traditional short-time average energy, short-time average zero crossing rate of technical feasibility demonstration of endpoint detection method the piano note recognition. Endpoint detection method based on the traditional method of using high frequency components, reducing the risk of leakage, accurate identification accuracy. Then the notes feature extraction, In combination with the acoustic characteristics of the piano music, from the frequency spectrum of the Fourier transform analysis notes, discusses the practical application of FFT zero padding. Get after detecting the peak of the spectrum information, the structure of the harmonic analysis. Process analysis method is given, as a result of this paper, the application of environment, no learning samples to help judging pitch, only by the judgement of twelve-tone equal temperament, this paper discusses to record all the same in the whole period of music notes first harmonic structure, then the harmonic structure of the overall harmonic characteristic of statistics to learn the piano.Systems using c # development, experimental cases there are both natural piano MIDI generated, playing the piano music style has a fast or slow, and the mixture of multiple parts, the results show that better effect can be obtained from the listening, the paper finally analyses the experimental results although note identification is feasible. As a whole in this paper, the notes recognition system collating and summarizing the research. Complete basic system design and development, and finally verified by the experimental design feasibility of the recognition system notes.