Recognition of Technical Action Types Based on Main Frequency of Hitting Sound Spectrum of Elite Badminton Players

Abstract

This article proposes a method to recognize badminton skills by using the sound of hitting the badminton. The authors record the six different sounds of badminton hits of world champion Lin Liwen. With the help of the open-source audio software Audacity, the sound spectrum of the six recorded sounds was analyzed. The main frequency of the sound spectrum is applied to recognize the technical action type. The results show that for different technical actions have. From high to low, the main spectrum frequency range represents the technical actions of smash, clear, drop, drive, lift, and net. The results mean that the main frequency of sound spectrum can reveal the type of technical action. Furthermore, a machine learning method, Fuzzy C-Means (FCM) algorithm is introduced in this paper to cluster the main sound spectrum frequency of different technical actions. The clustering result can clearly distinguish the technical action type, substantiating the feasibility of identifying the technical actions types by the main sound spectrum frequency. The proposed method can be applied to statistics of the technical actions during the badminton competition quickly. In addiction, it is convenient to recognize the technical actions during the training. Meanwhile, it is helpful to evaluate the teaching effectiveness during the teaching scientifically. Now everyone can use a smart phone to collect the sound of hitting the badminton and analyze the main sound spectrum frequency through free open-source software. Therefore, compared with installing expensive sensors and video equipment on badminton rackets, this proposed method is cost-saving and easily accessible to anyone with mobile devices. In addition, the proposed method is expected to become a universally applicable method for statistical analysis of badminton technical action.