An efficient algorithm of beat detecting and the implement on DSP

 The music beat denotes the organization of a fixed period of time with the strong beat and weak beat, which is also called as beat for short. The continuity of the music beat is expressed as the average speed of the music, and its unit is bpm(beat per minute).It is indispensable for digital turntable to detect the music beat and display the bpm on LCD screen, but the existing algorithms of beat detecting cost more runtimes. In the digital turntable, the DSP processor can not provide most of its resources on the beat detecting, which is not the main function of the digital turntable. A method is presented for searching the lowest sampling rate of beat detecting in a certain bpm extent, which is deduced from the rule of apperceiving music beat by human ears and observing the time-frequency characteristic of music, and then a very simple strategy of downsampling is presented. By extracting beat information from it, the principle and the method to estimate the bpm is proposed, which compose an efficient algorithm of beat detecting used in devices such as digital turntable. The principle of algorithm was described, as well as the process of implementation on a DSP processor and the testing results are also presented. Compared with traditional algorithms, due to without complex processing such as clustering detecting and multi-channel filtering and frequency characteristic extracting at higher sampling rate, so the proposed algorithm costs less runtime, so it is not necessary to use a higher speed DSP processor, the DSP processor can works on its other main tasks well. By experimenting on the digital turntable, It was verified that the accuracy rate is higher. So it can meet the needs of the playback equipments for real-time detecting beat.