The Design Of Music Analysis Sistyem Based On DSP

With the development of multimedia technique and expansion of the audio data, music analysis technique has been very important in music database retrieval, instrument tuning and auto notation. According to the description of MIDI format file, a melody can be identified by pitch and time values of the notes. At present, there are many music softwares like"Cakewalk"have integrated the function of music analysis. They can transfer the musical signal to notes in format of MIDI files by detecting the pitch and time length of the music notes. And then the musical signal can be saved in MIDI files. However, PC is necessary to use those softwares, which bring the majority of music lovers and creators a great deal of inconvenience. In this paper, a music analysis system based on DSP has been designed. It has the function of notation and musical instrument tuning.First, this paper provided a pitch recognition algorithm based on self-related processing and fast Fourier transform (FFT) by studying and researching the traditional pitch recognition algorithm (including parallel processing method, harmonic peaks method and wavelet analysis, etc.). And then the notes value recognition algorithm is designed according to the theory of Short-term technical analysis of the voice. The improved recognition algorithm is simulated. The simulation results show that the algorithm is feasible, fast and reliable.Second, a hardware system is designed by considering the factors of size, power consumption and real time. In the system, the DSP chip TMS320VC5502 is used as the core computing processors. The ARM chip S3C44B0X is used as main controller. The special low-power 24-bit analog-to-digital voice conversion chip CS53L21 is used for collecting audio signal. The detailed design of the hardware interface between the module and the bus connection is given.And then, this paper has carried on feasibility research on how to implement the recognition algorithm on the hardware system. C language is mainly used to design the program code. Compile language is used to design the Serious time-consuming algorithm such as FFT. So the real-time is greatly enhanced. The memory space is greatly saved by using queue structure to receive the sample data.Finally, a summary of this article is made. Some problems needs to be improved in the future works are given. And the new research trends in recent years are briefly introduced.