Application of Dance Rhythm Analysis and Music Matching Algorithm in the Choreography Process

Abstract

Improvised dancing choreographies represent a crucial area of research within cross-modal analysis. Central to this endeavour is the challenge of effectively correlating using a statistical one-to-many mapping, music and dance. This mapping is instrumental in generating authentic dances across diverse genres. In this manuscript, application of dance rhythm analysis and music matching algorithm in the choreography process (DRA-MMAC-SPGAN) is projected. The pictures are gathered from AIST++ Dance Motion information set are given as input. The input images are fed to pre-processing using Sub Aperture Keystone Transform Matched Filtering (SAKTMF) for remove the background sound from the input pictures. Afterward a pre-processed picture is provided to Holistic Dynamic Frequency Transformer (HDFT) for extracting the music characteristics like onset strength envelope, Mel-frequency cepstral constants, Chroma energy normalized and peak of onset strength envelope. Then the extracted features are given to MCoCo for segmenting the music beats. In general, Semantic-Preserved Generative Adversarial Network (SPGAN) does not discuss modifying optimization techniques to identify ideal parameters to guarantee accuracy dance generated based on music. Hence, the BFO is to optimize to SPGAN which accurately generate the dance choreography based on music. The proposed DRA-MMAC-SPGAN approach is applied in Python. The presentation of the suggested DRA-MMAC-SPGAN approach attains 22.54%, 26.36% and 25.95% higher accuracy, 20.63%, 23.86% and 25.96% higher recall and0.5%, 0.7% and 0.3% lower hit rate compared with existing methods like music-to-dance motion choreography with adversarial learning (MDMC-GAN), a deep music recommendation method based on human motion analysis (DMR-HMA-LSTM-AE) and generating dances with music beats using conditional generative adversarial networks (GDMB-CGAN) respectively.