Crime-Related Keyword Spotting in Low-Resource Nigerian Languages Using Knowledge-Distilled DSCNN for Security Monitoring

Insecurity remains a major challenge in Nigeria, creating a need for intelligent tools that can automatically detect crime-related keywords in recorded meetings and conversations. This paper presents a knowledge-distilled Depthwise Separable Convolutional Neural Network (D-DSCNN) system for keyword spotting in the three predominant Nigerian languages — Hausa, Igbo, and Yoruba. A custom dataset of crime-related keywords was collected from native speakers and expanded through data augmentation. The proposed system achieves 96.40% test accuracy with extremely low false positive and false negative rates, while maintaining a small computational footprint suitable for edge devices. The results demonstrate the practical feasibility of deploying efficient multilingual KWS systems for security monitoring in low-resource African language settings.

Keyword spotting, Nigerian languages, Hausa, Igbo, Yoruba, Knowledge distillation, Security monitoring, Low-resource speech, Edge devices

1. Inuwa-Dutse, I. (2025). NaijaNLP: A survey of Nigerian low-resource languages. arXiv preprint arXiv:2502.19784. [Google Scholar] [Crossref]

2. Kavya, H. P., and Venna, K. (2014). Sensitive keyword spotting for crime analysis. In Proceedings of the IEEE National Conference on Communication, Signal Processing and Networking (NCCSN), pp. 1–6. [Google Scholar] [Crossref]

3. Lei, L., Yuan, G., Zhang, T., and Yu, H. (2022). Low-power feature-attention Chinese keyword spotting framework with distillation learning. ACM Transactions on Asian and Low-Resource Language Information Processing, 22(2), Article 56. [Google Scholar] [Crossref]

4. Lepak, Ł., Radzikowski, K., Nowak, R., and Piczak, K. J. (2021). Generalisation gap of keyword spotters in a cross-speaker low-resource scenario. Sensors, 21(24), 8313. [Google Scholar] [Crossref]

5. López-Espejo, I., Tan, Z.-H., Hansen, J. H. L., and Jensen, J. (2021). Deep spoken keyword spotting: An overview. IEEE Access, 10, 4169–4199. [Google Scholar] [Crossref]

6. Sørensen, P. M., Epp, B., and May, T. (2020). A depthwise separable convolutional neural network for keyword spotting on an embedded system. EURASIP Journal on Audio, Speech, and Music Processing, 2020(1), Article 10. [Google Scholar] [Crossref]

7. Tan, Z.-H., Wang, J., Zhang, D., and Jensen, S. H. (2021). Knowledge distillation for small-footprint keyword spotting. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 29, 2730–2741. [Google Scholar] [Crossref]

8. Xie, J., Zhang, S., Hu, Y., and Wu, X. (2023). Multilingual and cross-lingual keyword spotting with pretrained speech models. arXiv preprint arXiv:2304.09585. [Google Scholar] [Crossref]

9. Zhang, Y., Suda, N., Lai, L., and Chandra, V. (2017). Hello Edge: Keyword spotting on microcontrollers. arXiv preprint arXiv:1711.07128. [Google Scholar] [Crossref]

• Evaluating the Impacts of Mind Mapping Strategy on Developing EFL Students’ Critical Reading Skills
• Significance of Reading Instructions for Language Improvement in Children with Down Syndrome
• Prenasalised Consonants in Liangmai
• Metadiscourse Matters: Definitions, Models, and Advantages for ESL/ EFL Writing
• Blank Minds and Stuck Voices: Understanding and Addressing Cognitive Anxiety in High-Stakes ESL Speaking Tests