Bhagya Shree
Department of Computer Science and Applications, Panjab University, Chandigarh, India.

Sonal Chawla
Department of Computer Science and Applications, Panjab University, Chandigarh, India.

DOI https://doi.org/10.33889/IJMEMS.2026.11.5.089

  ;

Abstract

Speech command recognition is an essential subsystem for Speech to Sign Language translation systems. It facilitates communication between the hearing disabled community and society. Researchers have been exploring speech command recognition techniques to improve the accuracy and efficiency so that these can be beneficial for hearing disabled persons. This research introduces a hybrid deep learning framework named Convolutional Residual Dense Network (CResD-Net) for efficient speech command recognition. This research pursues four key objectives. First, it reviews and compares the existing speech command recognition frameworks. Second, it presents an algorithmic framework leveraging CResD-Net to enhance speech command recognition efficiency. Third, the framework is evaluated through two research studies. In Research Study 1, an ablation analysis was carried out to understand the role of each component of CResD-Net. In Research Study 2, the proposed framework was compared with existing studies from literature to evaluate its performance. Various metrics like accuracy, WER, Weighted Precision, Weighted Recall and Weighted F1-score were reported in this experiment. Finally, the outcomes of the study were analyzed to identify their implications for improving speech command recognition systems.

Keywords- Residual network, Dense neural network, Speech command recognition, Convolutional neural network.

Citation

Shree, B., & Chawla, S. (2026). Algorithmic Framework for Speech Command Recognition Using Hybrid Deep Learning Models. International Journal of Mathematical, Engineering and Management Sciences, 11(5), 2228-2247. https://doi.org/10.33889/IJMEMS.2026.11.5.089.