Hybrid soft computing technique for music emotion recognition based on deep feature extraction and optimal feature selection with F CNN+ELNN

Authors

  • Dr. R. Sundar Author
  • Kavibharath L Author
  • Dr. R. Hema Author
  • G. Manikandan Author
  • Dr. M. Balakrishnan Author
  • Dr.N.Senthil Madasamy Author
  • Dr. Revathy P Author

Keywords:

music emotion, soft computing, cluster-driven, deep feature extraction, optimal feature selection

Abstract

Music holds a significant place in human history, serving as a means to convey emotions. With the growing importance of music emotion recognition in digital entertainment and human-machine interaction, there has been considerable advancement in this field. However, existing research often relies on short-term artificial stimuli, limiting their ability to evoke complex emotions and capture dynamic brain responses. While previous studies have segmented music comprehensively at similar intervals to analyze mood, recent efforts focus on recognizing mood by extracting sections with similar acoustic features. In our study, we propose a hybrid soft computing approach for music emotion recognition, integrating deep feature extraction and optimal feature selection techniques. Initially, we employ a feeding convolutional neural network (F-CNN) layer to extract deep features from audio signals. Subsequently, we introduce a modified wolf pack (MWP) cluster-based algorithm to optimize feature clustering, addressing data dimensionality challenges effectively. Additionally, we utilize an enhanced Laguerre neural network (ELNN) model for predicting emotional dimensions' levels. Our proposed method is evaluated using the Soundtrack dataset comprising 360 sound samples. The F-CNN+ELNN model achieves an impressive maximum accuracy of 97.568%, outperforming existing models by 15.689%. It signifies a significant advancement in music emotion recognition, demonstrating the efficacy of our approach in accurately capturing and interpreting emotional nuances in music.

Downloads

Download data is not yet available.

Downloads

Published

2026-09-22

How to Cite

Hybrid soft computing technique for music emotion recognition based on deep feature extraction and optimal feature selection with F CNN+ELNN . (2026). Minimax Theory and Its Applications, 11(2s), 544-559. https://journalmta.com/index.php/jmta/article/view/435