Buzzer audio analysis technique based on fast Fourier transform algorithm
In order to deeply understand and control the performance and production of buzzer sound effect,the research designed the audio analysis algorithm of buzzer based on the fast Fourier change,and firstly,the fast Fourier change and the improved LeNet-5 network were used to achieve the recognition and classification of the sound spectrum feature map.Simulation experiments show that,verified by the relationship coefficients,Euclidean distance and cosine distance indicators,the research improved Fourier-conv-olutional network model on the field-programmable gate array has a high degree of agreement with the results of the computer calcula-tions,and the computation time is reduced by 55.97％.Compared with other audio recognition algorithms,the research-designed al-gorithm performs better in the average absolute value error and root mean square error,taking the smallest value of 0.587 and 0.561,respectively,with a better loss effect and a better match with the real results.The buzzer audio analysis algorithm designed in this re-search based on fast Fourier variation can effectively achieve the identification and classification of audio events,which is of practical value for achieving the interaction of buzzer music playing.
NETL
NSTL
万方数据
