CENN: Capsule-Enhanced Neural Network with Innovative Metrics for Robust Speech Emotion Recognition

📝 Abstract

This paper introduces a groundbreaking Capsule-enhanced neural network (CENN) that significantly advances the state of speech emotion recognition (SER) through a robust and reproducible deep learning framework. By integrating capsule layers into the network architecture, CENN captures hierarchical part-whole relationships in emotional speech features that conventional neural networks fail to model effectively. The paper also introduces innovative evaluation metrics that provide a more comprehensive and clinically meaningful assessment of SER system performance, revealing performance dimensions not captured by standard accuracy and F1 measures.

📋 BibTeX Citation

@article{zhang2024cenn,
  title     = {{CENN}: Capsule-Enhanced Neural Network with 
               Innovative Metrics for Robust Speech Emotion 
               Recognition},
  author    = {Zhang, Huiyun and Huang, Heming and Zhao, Puyang 
               and Zhu, Xiaojun and Yu, Zhenbao},
  journal   = {Knowledge-Based Systems},
  year      = {2024},
  month     = {sep},
  publisher = {Elsevier},
  url       = {https://www.sciencedirect.com/science/article/pii/S095070512401133X}
}

🔗 Related Publications