The Research Of The Music Beat Extraction Based On Low-Pass Gaussian Filter

Despite the rising demand of the musical fountain practical application, the existing products are still in track musical fountain a single, fixed state, not flexible to change according to actual demand in the form of a fountain. To change the status quo, it's necessary to rely on computers to analyze music signals and extract the variation of the music. Rhythm is the pulse of the music, also the analysis, research and application foundation of the music. The beat is the most obvious manifestation of the rhythm, therefore the beat extraction is the key to grasp the musical variation.This paper summarizes the results of previous studies seriously, proposes a FIR digital low-pass filter based on Gaussian function to the beat extraction in allusion to the characteristics for the audio signal and the limitations of current extraction methods. Research content and the steps include the following:(1) According to characteristics of the beat in the music knowledge and reflection of signal processing, confirm the starting point of this study is to analyze the energy peaks in the audio signal.(2) Has the effect of high-fidelity PCM encoded audio signal as the object of study, preprocessing to eliminate its interference and less computation.(3) In allusion to PCM encoded audio signals with time-varying linear and non-steady-state characteristics, this paper uses FIR digital filter and short-time Fourier transform. Through the demonstration, the Gaussian window function is the best subjects for this window function because of its smallest feature bandwidth. Construct low-pass filter based on Gaussian, and process preprocessed signal to obtain the low-frequency energy envelope.(4) Sort the peak energy of the filtered results, then took the method of the beat interval estimation, combine with the characteristics of music rhythm excluding the peak does not meet the requirements to retain effective peak, get beat specific time point information.Experiments show that the method proposed in this paper has the accuracy and effectiveness in the application of the beat extraction. Meanwhile, analysis the shortage of other filtering from theory for audio signal, and the experimental results based on these filters and filter used in this paper were analyzed and compared, demonstrated that the proposed method, compared to other similar rhythm method, has a significant advantage.