Research On Recognition And Recording Of Musical Signals

As a fundamental part of musical signals recognition and music information recording and retrieval,the identification and estimation of musical signals and the retrieval of musical signals are increasingly concerned by relevant researchers.The diversity of musical signals determines the complexity and difficulty of their recognition.In order to solve the problem in musical signals recognition,this paper improves the analysis method of the signals in speech and music recognition.It mainly does the following work:According to the knowledge of Auditory Scene Analysis,combined with the knowledge of signal processing,music theory and music acoustics,the concept of separation and reconstruction of auditory elements is proposed.In order to solve the problem of large information redundancy in the current tone identification,this method divides the continuous frequency domain of the tone signal frame into rasterized frequency domain,and converts it into a reconstructed spectrum chart,harmonic distribution matrix and wave energy matrix to reduce the data of musical signal.The separation and reconstruction of auditory elements are used to improve the pitch recognition of musical signals in view of the inaccuracy in recognition of fundamental frequency loss,discordant tones,and low signal-noise ratio.The improved method can realize the accurate identification of many special cases such as harmonic loss,disharmony factor and low signal-noise ratio in musical tone recognition.By optimizing the calculation,the original method is improved,and the harmonic correction is used in this method to correct the information of fundamental frequency,solving the recognition error caused by insufficient resolution effectively,and improve the resolution ability successfully.The improved pitch recognition method can achieve 91%recognition accuracy of various pitch in the entire pitch area.For noise interference,it can tolerate a low signal-noise ratio of at least 13d B.For the identification of inharmonic tones,the maximum inconsistency factor of inclusiveness gets?_lim(28)0.004.For the problem that the resolution in the time-frequency domain cannot be taken into account at the same time,without increasing the amount of calculation,the harmonics of the redundant position of the resolution are used to correct the fundamental frequency harmonics to improve the static resolution.It can realize a 4-time improvement in the bass region and midrange region.At the same time,it is effective in the case of harmonic loss such as fundamental frequency loss,higher harmonic loss,etc.,and it can improve the accuracy of the recognition of fundamental frequency lost from the previous 0.3 to 0.85.In the case that the time domain resolution can meet the needs of daily identification,it has better real-time performance than the wavelet transform with the same minimum resolution,and the computational complexity is 0.01-0.05times of the wavelet transform.The improved method of musical signals'recording and retrieving can effectively achieve the matching of information recording and information of frames,and the storage of musical data can save 12%of the storage space.